This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Synthesis and Chemistry of Carbocyclic Mimics of β-Lactams as
lipopeptides, macrolides, polyethers, tetracyclines, and fluoroquinolones (Figure 1).2 Among
them, the β-lactam antibiotics are one of the most important classes in the world, which
account for more than half of the prescriptions for antibiotics worldwide.3, 4 The β-lactam
antibiotics can be used to treat both Gram-positive and Gram-negative bacterial infections.
They target bacterial enzymes called Penicillin-Binding Proteins (PBPs) that perform key
functions in the biosynthesis and remodeling of the peptidoglycan structure of the bacterial
cell wall,5 which is a process that is unique to bacteria. β-Lactams are more effective than
other antibiotics that aim at the intracellular substances since PBPs are more easily accessed.6
2
Figure 1. Chemical structures for some examples of major classes of antibiotics
The widespread use and the unfortunate abuse of antibiotics stimulate the evolution of
bacteria,7 many of which have developed resistance, by various mechanisms, towards every
major class of existing antibiotics, particularly the β-lactams. As early as the 1980s,
methicillin-resistant Staphylococcus aureus (MRSA) was noted in the United States. This
resistance became a serious problem because MRSA could resist all the β-lactams. With the
wide emergence of MRSA, the usage of vancomycin was significantly increased, since
vancomycin was the only antibiotic available at that time to effectively treat MRSA infection.
This inevitably led to vancomycin-resistant S. aureus (VRSA) and vancomycin-resistant
Enterococcus spp. (VRE), and this severe vancomycin-resistance problem eventually
developed into a hospital crisis by 1992.8 Obviously, new antibiotics with improved antibiotic
activity are urgently needed in order to limit bacterial resistance and maintain the effectiveness
of antibiotics.9
Generally, there are four fundamental mechanisms of bacterial resistance to antibiotics,
including enzymatic degradation of antibiotics, alteration of the antibiotic targets, decrease of
3
membrane permeability to antibiotics, and efflux.10 For β-lactam antibiotics, the most
significant bacterial resistance is caused by β-lactamases that can hydrolyze β-lactams,
especially in Gram-negative bacteria. β-Lactamases consist of two types, the serine β-
lactamases (SBLs, which are further grouped as Class A, C and D) and the metallo β-
lactamases (MBLs, also known as the Class B β-lactamases). More details about the β-
lactamases will be provided in Section 1.3. Another important resistance mechanism is that
some penicillin-binding proteins (PBPs) could alter the β-lactam targets, particularly in Gram-
positive bacteria.11
1.1 β-Lactam Antibiotics
All β-lactam antibiotics possess an essential four-membered amide (lactam) ring,
which are commonly fused with a five or a six-membered ring (except the monobactams). In
general, β-lactam antibiotics are classified into five groups depending on the difference of the
non-lactam rings, including penicillins, cephalosporins, carbapenems, penems and
monobactams, and some of the typical examples are shown in Figure 2.
Figure 2. Structures of some typical β-lactam antibiotics
4
1.1.1 Penicillins
On September 28, 1928, the Scottish bacteriologist Alexander Fleming accidently
discovered a bluish-greenish mould from the Petri dishes left on the bench for several days at
the Saint Mary's Hospital in London. This mould, later named as Penicillium notatum, was
found to be able to produce a liquid substance that could inhibit the growth of Staphylococci
and it was named penicillin.12 In 1929, Fleming reported that penicillin had potential
therapeutic application, since it had great activity against several human pathogenic species
including Staphylococcus, Streptococcus, Pneumococcus and Gonococcus. More importantly,
penicillin showed low toxicity to experimental animals such as mice and rabbits.13
No significant research progress on penicillin was made until 1938, when a German-
born British biochemist Ernst Chain was invited by Sir Howard Florey to join his group in the
Sir William Dunn School of Pathology at the University of Oxford, where they began to study
some naturally produced antibacterial substances including penicillin.14 In 1941, an advanced
procedure for purifying penicillin was developed by Norman Heatley of the Florey group.
Extraction of crude penicillin by this method and further purification provided a 50% pure
penicillin at the end of 1941.15, 16
In 1942, the Florey group started to collaborate with Sir Robert Robinson at the Dyson
Perrins Lab at Oxford to determine the structure of penicillin, which was just known as a
carboxylic acid at the time.16 The structural determination focused on the degradation products
of penicillin such as penicillamine and 2-pentenylpenilloaldehyde (Figure 3). Later in 1943, it
was discovered by Dorothy Hodgkin and Wilson Baker that penicillin contained C, H, N, O
and S through elemental analysis of the penicillaminic acid and penicillamine.16
5
Figure 3. Structures of degradation products of penicillin F
In August 1943, US researchers Wintersteiner, MacPhillamy and Alicino at Squibb,
obtained a crystal of the sodium salt of penicillin and they reported a molecular formula of
C16H17O4N2SNa.16 Following this discovery, the Florey group crystallized a barium salt of
penicillin that provided another formula of (C14H19O4N2S)2Ba.16 Later on, the American
penicillin was found to contain a benzyl group and named as penicillin G while the Oxford
penicillin was named as penicillin F that contained a Δ2-pentenyl group.17
Right after the confirmation of penicillin's molecular formula, Robinson found that
penicillin could be hydrolyzed to penicilloic acid under alkali conditions, and then he proposed
a thiazolidine-oxazolone structure (Figure 3) for the Oxford penicillin. He also assigned a
correct structure for an isomeric penillic acid possessing two acid groups and an imidazoline
ring, which was formed from penicillin at pH 2 (Figure 3).16,17 From then on, several different
structures were proposed for penicillin. The β-lactam structure proposed by Abraham and
Chain in 1943 was indeed the correct one. The debate on the structure of penicillin did not stop
until 1945, when the structure was unambiguously confirmed by X-ray crystallographic
analysis by Crowfoot and Low at Oxford, which demonstrated the presence of a β-lactam ring
from the crystals of the Na, K and Rb salts of penicillin G.18
The discovery of penicillin created a new era for antibiotics and its clinical use has
6
saved numerous lives, and it became one of the most important antibiotics in human history.
Sir Alexander Fleming, Ernst Chain and Howard Florey were awarded the Nobel Prize in the
field of medicine in 1945 for their contributions to the discovery and application of penicillin.
The chemical synthesis of penicillin was extremely difficult even after its structure was
determined. Large scale production of penicillin still relied on fermentation methods until
1957, when the first reasonable total synthesis of penicillin was reported by Sheehan that
offered penicillin V in about 10% yield.19 Nowadays, the majority of clinical penicillins such
as ampicillin, methicillin, oxacillin and so forth, are obtained by semi-synthesis starting from
6-aminopenicilllanic acid (6-APA), which is produced in industrial scale fermentations of
Penicillium chrysogenum.20
Scheme 1. Preparation of 6-APA and semi-synthetic penicillins
1.1.2 Cephalosporins
In 1945, Giuseppe Brotzu, an Italian professor studying hygiene at the University of
Cagliari, noticed that the young people who regularly swam at Su Siccu Bay, near the end of
the city sewer system, were less likely to suffer from typhoid fever.12 Brotzu tested the sewer
water sample of the bay and found that it could inhibit the growth of Salmonella typhi culture.
Later, he discovered that some substance produced by a fungus in the sewer water had potent
activity against Gram-negative bacteria. The fungus was isolated and named Cephalosporium
acremonium. Further studies showed that this fungus had broad-spectrum antibacterial activity
and Brotzu then sent a sample of this organism to the Florey group at Oxford.21
7
In Florey's lab, the first antibiotic extracted from Cephalosporium acremonium was
named cephalosporin P, which showed activity against certain Gram-positive bacteria. In 1951,
a second compound extracted from this organism was named as cephalosporin N (also known
as penicillin N), which had broad spectrum activity against both Gram-positive and Gram-
negative bacteria. Unfortunately, the bioactivity of cephalosporin N was significantly
decreased in the presence of a penicillinase (a β-lactamase) that was produced by Bacillus
subtilis.12 In 1953, a third compound was extracted from Cephalosporium acremonium by Guy
Newton and Edward Abraham, which was then named cephalosporin C and which possessed a
broad spectrum activity against many Staphylococus aureus strains.12 Later on, it was found
that this compound presented greater resistance to β-lactamases and it saved mice infected by
penicillin-resistant Staphylococci.12,21 The structure of cephalosporin C was determined by the
Oxford chemical crystallography lab of Dorothy Crowfoot Hodgkin in 1961.22
Because of the significant bioactivities of cephalosporin C, a large number of chemists
were involved in its total synthesis. The first total synthesis of cephalosporin C was published
by Woodward in 1966. At present, cephalosporin C is mainly produced by fermentation of
Acremonium chrysogenum.23,24 Upon treatment with dilute acid (Scheme 2), it is converted
into 7-aminocephalosporanic acid (7-ACA),25 which is the essential intermediate for producing
various cephalosporin derivatives. Thus, cephalosporin C became the prime starting point to
develop several generations of (semi-synthetic) cephalosporins.
Scheme 2. Cephalosporins N, C and preparation of 7-ACA
8
Based on their time of development and biological characteristics, cephalosporins have
been classified into five generations so far. The first generation of cephalosporins was
developed in the mid-1960s and included cephalothin, cephaloridine, cefazolin and others,
which possess simple 7β-acylamino side chains. The antibiotics of this family have a decent
capacity to eliminate Gram-positive pathogens such as Staphylococci and Streptococci.
However, their ability to fight against Gram-negative bacteria is modest.21
The second generation cephalosporins were developed between the late 1970s and
early 1980s. In these cephalosporins, the 7β-acylamino side chain was significantly modified
and the 3' substituent was altered as well. This group of cephalosporins, including
cefamandole, cefaclor and ceftazidime and others, have slightly less activity against Gram-
positive bacteria when compared to the first generation. However, they have better stability
towards β-lactamases and are more efficient at eliminating Gram-negative bacteria.17,21
In the 1980s, the third generation cephalosporins were introduced into clinical use.
Generally, they are more expensive than the previous two generations, but they are extended-
spectrum antibiotics with much better resistance to β-lactamases. Members of this category
(e.g. cefotaxime, cefixime and ceftbuten) possess a 7β-(aminothiazoyl)-oxyiminoacetamido
moiety, which plays a crucial role in their resistance to β-lactamases, especially the class A
9
SBL. Compared to the first two generations, the third generation has slightly less activity
against Gram-positive bacteria, but greater in vitro activity against Gram-negative aerobes.21
Cefepime and cefpirome are the typical examples of the fourth generation
cephalosporins, which were introduced clinically in 1990s. They have extended Gram-
negative coverage and increased resistance towards β-lactamases. Due to the positive charge in
the 3' quaternary ammonium side chains, their ability to inhibit Class C SBL was
increased.17,21
Ceftobiprole and ceftaroline are the fifth generation of cephalosporins, which are still
under development, and both of them have potent activity against MRSA. Ceftaroline has
extended spectrum activity against Gram-negative pathogens, while ceftobiprole possesses
wide spectrum coverage of both Gram-positive and Gram-negative bacteria.17,26
10
1.1.3 Carbapenems
β-Lactam antibiotics were greatly threatened by bacterial β-lactamases starting from
the late 1960s. Therefore, a search for new β-lactam antibiotics and β-lactamase inhibitors
became increasingly important.27 At the moment, the Beecham Laboratory was a leader in
research on β-lactamase inhibitors, and in 1976 researchers at this company discovered the
first β-lactamase inhibitor, olivanic acid, which was produced by the Gram-positive bacteria
Streptomyces clavuligerus.27 Olivanic acid possesses a carbapenem backbone, which is a four-
membered β-lactam ring fused with an unsaturated five-membered ring. Olivanic acid works
as a broad-spectrum β-lactam antibiotic, however, it is not used clinically since it is not stable
and cannot readily penetrate the bacterial cell.27 In the meantime, scientists at Merck
discovered thienamycin from the fermentation broth of Streptomyces cattleya.21 Compared to
penicillins and cephalosporins, the sulphur atom is replaced by a carbon atom in carbapenems.
Thienamycin has a hydroxyethyl substituent at C6 rather than the acylamino group, which
enables thienamycin to have remarkable activity against both Gram-positive and Gram-
negative bacteria. In addition, the stereochemistry of C5, C6 and C8 has essential influence on
the corresponding stability and bioactivity. It was found that the thienamycin with trans
stereochemistry between C5 and C6 and a (R) stereochemistry at C8 presented the greatest
stability and antibacterial activity.21,28
In 1978, the Merck research group published the first total synthesis of thienamycin
followed by a modified synthetic process developed by the same lab two years later.21
Unfortunately, it was discovered that thienamycin is not stable in aqueous solution and is
11
sensitive to mild base (above pH 8). Even worse, it is highly reactive to nucleophiles including
hydroxylamine and even its own primary amine.27 In 1979, the Merck group noticed that the
conversion of the primary amine to an N-formimidoyl group could relieve the instability of
thienamycin in aqueous solution. The new antibiotic was named imipenem and showed high
affinity for PBPs and higher stability towards β-lactamases. Unfortunately, imipenem was
found to be the substrate for human renal dehydropeptidase I (DHP-I); therefore, the DHP-I
inhibitor cilastatin was developed, to be used in combination with imipenem in clinic.21
In the following 20 years, other clinically useful carbapenems, including meropenem,
ertapenem and doripenem were developed. The presence of a methyl group at the C1-β
position in these carbapenems makes them resistant towards DHP-I, thus eliminating the need
of a DHP-I inhibitor.27
Among the recently developed carbapenems, sanfetrinem by Glaxo Wellcome was an
unusual one as its structure consists of three rings, and it is highly stable towards human renal
DHP-I and β-lactamases while possessing broad-spectrum activity against both Gram-positive
and Gram-negative bacteria.21 Carbapenem PZ-601 was discovered at Sumitomo
12
Pharmaceuticals and is still in clinical trials, aimed at treatment of some serious infections
caused by MRSA and extend-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae.21
1.1.4 Penems
Penems were entirely obtained through total synthesis rather than bacterial
fermentation.29 The very first synthesized penem was 7-phenyl-2-methyl penem, prepared by
the Woodward group in 1976. The penems possess a core skeleton that is a hybrid of that of
the penams (penicillins) and cephems (cephalosporins).30
Based on this core structure, a number of penems such as SCH 29482, HR 664 and
sulopenem, has been synthesized. Unfortunately, only a few penems (e.g. sulopenem,
ritipenem and faropenem) are used clinically due to their general instability and toxicity. In
general, penems are classified into five groups (A-E) depending on the side chains at the C2
position.30 Group A penems are thiopenems including SCH 29482, SCH 33343 and sulopenem.
SCH 29482 has potent activity and SCH 33343 is undergoing Phase III clinical trials.
Sulopenem, discovered in the 1980s, possesses good activity against both Gram-positive and
Gram-negative excluding Pseudomonas aeruginosa.29 However, sulopenem was not used
13
clinically until a pro-drug (PF-03709270) was discovered, which could be used for the
treatment of community-acquired pneumonias.21
Oxypenems are Group B penems, which include HR 664 that is the most active
compound in this family.30 Alkylpenems belong to Group C and ritipenem is a typical example
of this category. Ritipenem was developed and introduced by Tanabe Seiyaky Co in Japan,21
and shows good activity against a wide range of bacteria including Enterobacteriaceae,
Haemophilus influenzae, and S. aureus.29 Arylpenems and faropenem are representatives of
the Group D penems. The latter was developed by Daiichi Asubio Pharma in Japan and
became commercially available in 1997,21 which not only has broad spectrum antimicrobial
activity but also relatively high stability towards the SBLs.17 Group E penems are aminopenem
derivatives, for example 2-aminomethyl penem, which resist hydrolysis by DHP-I due to the
presence of the 2-aminomethyl side chain. The aminomethyl group reduces its activity against
Gram-negative bacteria, but an unsaturated N-heterocyclic derivative such as one bearing an
imidazole group has enhanced antibacterial activity.30
In summary, the clinically useful penems demonstrate a broad spectrum of antibacterial
activity; however, they are not active against MRSA, Enterococci and P. aeruginosa.
14
Compared to imipenem (a carbapenem), penems have similar stability towards β-lactamases,
but they are more stable towards DHP-I.21
1.1.5 Monobactams
In the 1970s, many research groups worldwide were searching for β-lactamase
inhibitors. In 1976, a Japanese group at the Tekeda Company discovered the first monobactam
(i.e. a monocyclic β-lactam) called nocardicin A, which was produced by an organism later
named as Nocardia uniformis tsuyamanensis. However, nocardicin A has only moderate in
vitro activity against Gram-negative pathogens such as Proteus and Pseudomonas.21 Later on,
a group of modified monobactams were discovered from the fermentation broth of Flexibacter
alginoliquefaciens YK-49, and those possessing a 3α-formylamino substituent were named as
formadicins (A-D). Even though formadicins have a narrow spectrum of antimicrobial activity,
they are stable towards β-lactamases due to the 3α-formylamino substituent.21
In the early 1980s, a series of monobactams including sulfazecin and isosulfazecin
were found by the Tekeda group. Meanwhile, the Squibb team discovered SQ 26180 and SQ
26970 produced by Agrobacterium radiobacter.21 These monobactams have activity only
against Gram-negative bacteria.31
15
In 1983, aztreonam was successfully synthesized by chemists at Squibb. It showed
potent activities against Gram-negative pathogens. As a result, aztreonam has been used in
clinical treatment of infection caused by Gram-negative bacteria since 1986, but it is
inactivated by Gram-positive bacteria. Later on, another monobactam named carumonam was
discovered by Takeda,21 and was used for treating Gram-negative bacterial infections.
Carumonam and aztreonam have similar spectra of antibacterial activity and stability towards
β-lactamases. They are inhibitors of Class C SBLs, but are inactivated by Class A ESBLs.
Surprisingly, although they were still hydrolyzed by MBLs, the rate was fairly slow,21 which
makes them potential MBLs inhibitors.
Compared to carumonam and aztreonam, oximonam and tigemonam have similar
bioactivity and stability towards β-lactamases, but very little antimicrobial activity against
Gram-positive bacteria.21 BAL 30072 is an experimental monobactam with a siderophore
moiety that improves uptake by P. aeruginosa via an active iron uptake system. BAL 30072 is
stable to MBLs and has been examined against multidrug-resistant Gram-negative
bacteria.21,32
16
1.2 Penicillin Binding Proteins
Between the 1940s and 1970s, much research was devoted to the analysis of the
structure of the bacterial cell wall, in order to understand the mechanism-of-action of β-
lactams. In the 1970s, it was discovered that penicillin interacts with several target enzymes
that are involved in the synthesis and maintenance of the bacterial cell wall, which were
consequently named penicillin-binding proteins (PBPs).17
Penicillin binding proteins could be divided into two primary groups based on their
molecular weight, the high molecular weight (HMW) PBPs (MW > 55000) and the low
molecular weight (LMW) PBPs.33 HMW PBPs are multimodular and are involved in the
polymerization of peptidoglycan and insertion into pre-existing cell well, while the LMW
PBPs are responsible for endopeptidase, transpeptidase and carboxypeptidase activities.34
According to their different functions, each main group of PBPs could be further categorized
into three subclasses (Class A, B and C). The Class A HWM PBPs (e.g. PBP1a and PBP1b of
E.coli) catalyze the elongation of uncross-linked glycan chains because of their N-terminal
domain that controls their glycosyltransferase activity.34 HMW PBPs in class B (e.g. PBP2 and
PBP3 of E. coli) are responsible for cell morphogensis.34 The Class C HWM PBPs are
considered to be penicillin sensory proteins, such as BlaR, which is responsible for inducing
the synthesis of β-lactamases.33 In terms of cell activities, the three classes of LMW PBPs are
inferior to the HMW counterparts.
1.2.1 Peptidoglycan and D-Ala-D-Ala Transpeptidases
In the 1960s, the structure of bacterial cell peptidoglycan and its biosynthesis were
gradually elucidated. Schleifer and Kandler considered that bacterial cell was made up of
acylated amino sugars and three to six different amino acids, which were called by a few
17
different names including basal structure, mucopeptide, glycopeptide, glycosaminopeptide,
murein and peptidoglycan. Among these names, peptidoglycan is a more precise term to
describe the polymer of the bacterial cell wall.35 It was discovered that for both Gram-positive
and Gram-negative pathogens, the glycan polymer consists of alternating N-acetylglucosamine
(NAG) and N-acetylmuramic acid (NAM). In addition, there is a peptide unit attached to the
D-lactyl portion of NAM in order to connect the glycan polymers.35 More specifically, the
pentapeptide unit is [L-Ala–γ-D-Glu–m-DAP–D-Ala–D-Ala] in Gram-negative bacteria
(Scheme 3), while in Gram-positive bacteria the corresponding unit is [L-Ala–γ-D-Glx–L-
Lys(ε-Gly5)–D-Ala–D-Ala].35
Slaton's experiments in 1961 suggested that the bacterial cell wall was highly cross-
linked and this cross-linking process was very likely carried out during the final step of
peptidoglycan biosynthesis.36 In 1965, it was found by Wise and Park that penicillin could
inhibit the synthesis of bacterial mucopeptide through a cross-link reaction, which was a
transpeptidation reaction involving the loss of D-alanine. They also provided a hypothesis that
during the inhibiting process, penicillin was acting as a mimic of the L-Ala–γ-D-Glu segment
of the substrate.37 Later on, Tipper and Strominger proposed that penicillin was a
conformational analogue of the D-Ala-D-Ala in the linear glycopeptide instead of the L-Ala–γ-
D-Glu segment (Figure 4), which acted as an inhibitor of the peptidoglycan transpeptidase to
generate a stable penicilloyl enzyme intermediate (Scheme 3).38,39 As a result, the
transpeptidation process was prevented and the peptidoglycan could not be formed, leading to
the incomplete generation of bacterial cell wall. This hypothesis then was widely accepted and
considered as a primary milestone of understanding the penicillin mode-of-action.17
18
Figure 4. Structural analogy of penicillin and D-Ala-D-Ala fragment of peptidoglycan
Scheme 3. Reaction of PBPs in E.coli involving biosynthesis of the bacterial cell wall and the inactivation of PBPs by penicillin 1.2.2 PBP-based Resistance to β-Lactam Antibiotics
The PBPs are transpeptidases or carboxypeptidases playing an essential role in
peptidoglycan metabolism. Modification of PBPs can lead to PBPs with low affinity for the
antibiotics.40 These PBPs are still functional as transpeptidases, but do not readily bind to the
β-lactams (e.g. PBP1a and PBP2b of S. pneumoniae).41
19
1.3 β-Lactamases
β-Lactamases are enzymes produced by many bacteria, which can efficiently hydrolyze
β-lactam antibiotics. The first known β-lactamase, produced by a strain of Bacillus coli that
could inactivate penicillin was discovered by Abraham and Chain in 1940.42 In the next six
decades, with the discovery and clinical use of an increasing number of β-lactam antibiotics,
bacteria evolved to produce various β-lactamases in order to defend themselves against the β-
lactam antibiotics. With the appearance of ESBLs (extend-spectrum β-lactamases) and
carbapenemases, the current β-lactam antibiotics suffer serious threats.17
In 1973, Richmond and Sykes proposed the first widely accepted classification of β-
lactamases. According to their favoured substrates, β-lactamases could be divided into five
groups, including Class I (cephalosporinases), Class II (penicillinases), Class III broad-
spectrum enzymes that were resistant towards p-chloromercuribenzoate but sensitively
inhibited by cloxacillin, Class IV enzymes that possessed the opposite activity as the Class III
counterparts, and Class V penicillinases that could inactivate cloxacillin but were sensitive to
p-chloromercuribenzoate.43
In 1980, a novel classification of the β-lactamases was developed by Ambler, which is
based on their structural properties at the enzyme's active site. The Ambler rule divides the β-
lactamases into serine β-lactamases (SBLs) and metallo β-lactamases (MBLs) and the serine
β-lactamases are further divided into Class A, C and D while the metallo β-lactamases are
designated as Class B.21 All SBLs possess a crucial serine residue in their active sites, and the
MBLs require the present of a metal ion (Zn2+) in their active sites.
20
1.3.1 Ambler's Class A β-Lactamases
The Ambler's Class A SBLs include the penicillinases of S. aureus PC1, B.
licheniformis 794/C, B. cereus 569/H β-lactamases and the broad-spectrum β-lactamases
TEM-1 and TEM-2 and others. The active sites of penicillinases possess a serine residue that
is similar to that of D-alanine carboxypeptidases.44
TEM-1 was discovered in the early 1960s from a single strain of E. coli, and after a
few years it was widely found in various species including Enterobacteriaceae, P. aeruginosa
and others. TEM-1 production resulted in resistance towards ampicillin, penicillin and the
early cephalosporins.45 Later on, as the first mutation of TEM-1, TEM-2 was discovered with
an amino acid substitution that was potentially producing the ESBL phenotype. In 1989, it was
reported that TEM-3 firstly presented the ESBL phenotype.45 So far, more than 90 additional
TEM mutations have been found, some of which are resistant to β-lactamase inhibitors and
thus called inhibitor-resistant β-lactamases, while the majority of the new TEM mutations are
ESBLs.45
In the 1980s, oxyimino-cephalosporins were applied to treat infections caused by
Gram-negative bacteria that produce the TEM family β-lactamases. Inevitably, new β-
lactamases were discovered in Enterobacteriaceae, and P. aeruginosa, which are the SHV
family of ESBLs with the ability to efficiently hydrolyze the oxyimino-cephalosporins.45 The
SHV family is the second largest group in the Class A SBLs, and they have similar structures
to the TEM enzymes.17
In the mid-1980s, the most widespread β-lactamases CTX-M were initially reported.
Their rate of spread has significantly increased since 1995.46 The CTX-M family are plasmid-
mediated ESBLs and mainly found in strains of Salmonella enterica, but they were not related
21
to the TEM and SHV families. The CTX-M type β-lactamases preferentially hydrolyze
cephalothin, cephaloridine and cefotaxime.45
There are some Class A SBLs with carbapenemase activity including SME, IMI,
NMC-A, KPC and GES families, which have a broad spectrum activity against penicillins,
early cephalosporins, carbapenems and aztreonam.17
The structure of the Class A SBLs active site was determined by X-ray crystal studies
and site-specific mutagenesis in the early 1990s. All Class A SBLs have similar structures at
their active sites, which contain the conserved residues Ser70, Glu166 and Lys73, and an
oxyanion hole.47 For the hydrolysis mechanism by SBLs, at the beginning, it was proposed
that Glu166 acted as a general base that participated not only in the acylation process, but also
the deacylation process, and activated a Ser70 and hydrolytic water for each process. However,
in 1992, the Strynadka group discovered, through an X-ray crystallographic study of the acyl
enzyme intermediate that was generated from penicillin G bound to the TEM-1, that Lys73
acted as the general base to deprotonate Ser70 to furnish the nucleophile (Scheme 4).48 After
the nucleophilic attack, a tetrahedral intermediate was obtained and then converted into the
acyl enzyme intermediate. For the deacylation process, like the initial viewpoint, it was
discovered that Glu166 worked as a general base to activate a water molecule that then
nucleophilically attacked the penicilloyl carbonyl group.
22
Scheme 4. Proposed mechanism for hydrolysis of penicillins by Class A SBLs (TEM-1)49
So far, this mechanism of deacylation is widely accepted, while that of acylation is still
under debate. In 2002, the Shoichet team proposed that Glu166 acted as the general base to
activate the water molecule and the proton of Ser70 was acidified by Lys73 and then the
activated water deprotonated Ser70 to generate the nucleophile.50 Recent work by the
Mobashery lab suggested that Glu166 activated Lys73 first, Ser70 was then deprotonated by
an unprotonated Lys73, which could promote the formation of the acyl enzyme intermediate.51
1.3.2 Ambler's Class C β-Lactamases
The class C SBLs are also called AmpC cephalosporinases. In 1981, the
chromosomally mediate AmpC enzyme of E. coli K12 was identified by Jaruin and
Grundstrom, which did not have sequence homologies as the class A SBLs, thus, it was
considered as the first Class C SBLs.44 In the late 1980s, an increasing number of plasmid-
mediated AmpC β-lactamases was discovered. For instance, CMY-1, which could inactivate
23
cefoxitin and cefotetan as well as penicillins, was isolated from a K. pneumoniae in 1989.
Moreover, the BIL, FOX, MOX, LAT, ACT and DHA families of β-lactamases all belong to
the Class C SBLs.52 In general, these β-lactamases are less active towards penicillins, but more
active towards cephalosporins, oxyimino-cephalosporins and monobactams.53 In the mid-
1990s, the first extended-spectrum Class C SBL, which was able to hydrolyze oxyimino β-
lactam antibiotics was isolated clinically and called GC1.54
For the β-lactam hydrolysis mechanism, Class C SBLs are essentially the same as the
Class A SBLs, including the active site acylation and hydrolytic deacylation.55 However, there
are still two significant differences. First of all, the rate-determining step is different. For Class
A enzymes, it is the acylation step, while the rate-limiting step for Class C counterparts is the
deacylation process. Second, during the deacylation step, the hydrolytic water molecule
approaches the acyl intermediate through two opposite directions. In the case of Class A SBLs,
water approaches the α-face, whereas in Class C enzymes it approaches the β-direction.55 In
terms of the structure of the active site, similar to that of Class A SBLs, the active site of Class
C SBLs consists of Ser64, Lys67, Lys315 and Tyr150, which occupy similar positions as the
corresponding amino acids of Class A SBLs active site.17 Since the active site of Class C SBLs
contains Tyr150 instead of Glu166, Tyr150 is considered to act as the general base involved in
both the acylation and deacylation processes of hydrolyzing β-lactams.56
1.3.3 Ambler's Class D β-Lactamases
The Class D SBLs are oxacillinases (OXAs), and OXA-1, OXA-2 and OXA-10 are the
first Class D enzymes, discovered in late 1980s. The OXAs possess very similar structures to
each other, but are very different from that of Class A and C SBLs.57 During the first two
decades since their discovery, there were only 20 Class D SBLs. Unfortunately, this family
24
was expanded widely in the following ten years, and more than 250 members were found for
this family by 2013.58 The Class D SBLs are usually found among the clinically important
bacteria such as Acinetobacter baumannii, P. aeruginosa, E. coli and Proteus mirabilis, which
could inactivate cephalosporins, the combination of β-lactam/β-lactamase inhibitors, and
carbapenems. For example, the enzymes produced by A. baumannii (e.g. OXA-23, OXA-24
and OXA-58) result in resistance towards carbapenems. In addition, some β-lactamases in P.
aeruginosa (e.g. OXA-14, OXA-28 and OXA-35) can inactivate extended-spectrum
cephalosporins. Recently, a new member OXA-45 was discovered in Turkey. It is a
cabarpenemase produced by K. pneumoniae, leading to considerable death in Europe and now
detected in North America.58
The Class D SBLs share a similar serine-initiated hydrolysis mechanism as the Class A
and C SLBs.59 In contrast to the active site of Class A enzymes, that of OXAs lacks an
analogous glutamate acting as the general base.58 In 2000, the Strynadka and Mobashery
groups reported two X-ray crystal structures of OXA-10 and they proposed that the Lys70 was
the general base.60,61 Later on, the Mobashery lab hypothesized that the Lys70 was
carboxylated (Lcx70) to produce a carbamate, which was considered to act as the general base
in the acylation process. Lcx70 directly deprotonates Ser67 to generate the nucleophile that
then attacks the carbonyl of the β-lactams.62 In 2013, the Bonomo group confirmed
Mobashery's proposal (Scheme 5).58 For the deacylation step, Lcx70 works as the general base
as well, which activates the water molecule that then underwent a nucleophilic attack on the
penicilloyl carbonyl.
25
Scheme 5. Proposed mechanism for hydrolysis of penicillins by Class D SBLs (OXA-10)
1.3.4 Ambler's Class B β-Lactamases
The Class B β-lactamases are metallo β-lactamases. The first Class B enzyme isolated
from Bacillus cereus was named BcII in the mid-1950s, and was found to be capable of
hydrolyzing semi-synthetic penicillins and cephalosporins.17 MBLs were not considered a
problem for the β-lactam therapy back then, since they were chromosomally encoded enzymes
and were not found in clinically important bacteria. In the next decade, it was discovered that
the MBLs could be inhibited by metal chelators such as EDTA, but recovered upon the
addition of ZnSO4,14 indicating the essential role of Zn2+ (metal) with the MBLs. However,
nowadays the Class B MBLs demonstrate a broad spectrum resistance to virtually all β-lactam
antibiotics except the monobactams. Even worse, they can hydrolyze the mechanism-based β-
lactamase inhibitors including sulbactam and tazobactam (section 1.4.1).63
In the late 1990s, the MBLs were divided into three subclasses (B1, B2 and B3) based
on their primary amino acid sequence homology. The B1 subclass include the most common
MBLs worldwide including BcII, CcrA, BlaB and IMP, VIM, SPM, GIM and NDM
families.63 CcrA produced form Bacteroides fragilis was discovered in the 1980s. In the next
10 years, the plasmid encoded IMP and VIM families were found in Gram-negative bacteria
including P. aeruginosa, Enterobacteriaceae and A. baumannii.63 Both of them showed a
26
broad spectrum of activity against all β-lactams except aztreonam (a monobactam). In the
early 2000s, the SPMs and GIMs were isolated from P. aeruginosa in Brazil and Germany
respectively, and possessed similar spectra of activity as IMPs and VIMs. In 2008, a new MBL
named NDM-1 was detected in K. pneumonia and E. coli from a patient who was returning to
Sweden from India.63 NDM-1 then spread very rapidly and it could inactivate the majority of
β-lactams except colistin.17 Subclass B2 involves CphA, ImiS and Sfh-I, and B3 subgroup
includes L1, FEZ-1 and the GOB family. In 2010, Bush proposed a new classification of
MBLs depending on their functions. The MBLs could be classified into two subgroups (3a and
3b). The 3a subclass includes penicillinases and cephalosporinases while the 3b subgroup
contains carbapenemases.64
Since the mid-1970s, studies on the mechanism for hydrolysis of β-lactams by MBLs
started. It was discovered that there were two zinc binding sites present in the active site of
many MBLs. Between the 1980s and 1990s, it was reported that Class B1 and B3 MBLs
require two zincs in order to obtain maximal activity, while Class B2 MBLs need only one
zinc to bind with the β-lactams. B2 MBLs present a relative narrow spectrum activity when
compared to the B1 and B3 ones.17,63 In the mid-1990s, X-ray crystallographic studies
suggested that the Zn1 binding site of B1 MBLs contains three histidines, known as the 3H
site, and the Zn2 binding site called DCH site involves asparagine, cysteine, and histidine
amino acid residues.63 For B3 MBLs, their Zn1 binding sites have similar amino acids residues
to those of B1 MBLs, but the corresponding Zn2 sites of B3 enzymes consist of one
asparagine and two histidines, known as the DHH. As mono-zinc MBLs, B2 enzymes contain
a DCH Zn2 binding site similar to that of B1/B3.63
27
The proposed mechanism for B1 and B3 MBLs involves Zn1 acting as a Lewis acid
binding to the oxygen of the β-lactam moiety, which increases the electrophilicity of the
carbonyl carbon for nucleophilic attack.65 At the same time, the β-lactam ring nitrogen
interacts with the Zn2 site (Scheme 6).63 There is a bridging hydroxide that is oriented
properly by the Asp through forming an H-bond between the Zn1 and Zn2 sites. Then the
hydroxide between the zinc ions attacks the carbonyl to give the tetrahedral intermediate,
which then further generates an enzyme-bound intermediate with a negative charge on the
nitrogen atom that can be stabilized by the Zn2 site. 63, 65
Scheme 6. Proposed mechanism for cephalosporins hydrolysis by B1 and B3 MBLs
In terms of the mechanism for B2 MBLs (Scheme 7), only the Zn2 site binds with the
β-lactam and a water molecule is activated by the Asp of the Zn2 site and a His nearby, which
then attacks the carbonyl. The anionic intermediate is stabilized by Zn2 as well, followed by
further hydrolysis, the enzyme-bound intermediate is degraded to the hydrolyzed product and
releases the free enzyme.17, 63
28
Scheme 7. Proposed mechanism for carbapenems hydrolysis by B2 MBLs
1.4 β-Lactamase Inhibitors
The β-lactam antibiotics are seriously threatened by the β-lactamases, which is an
urgent problem that needs to be addressed. There are two feasible strategies to relieve this
situation, including discovery of novel antibiotics that are not the substrate of β-lactamases
and development of β-lactamase inhibitors to be used in combination with β-lactams, which
could prevent the β-lactams from being hydrolyzed by the β-lactamases.66
1.4.1 β-Lactams as β-Lactamase Inhibitors
The first example of β-lactamase inhibition was reported in 1956. Cephalosporin C was
resistant to penicillinase, therefore it was a potential inhibitor for certain penicillinases.
Abraham and Newton found that Cephalosporin C could competitively inhibit the hydrolysis
of penicillin G and N, because it was more sensitive to the S. aureus penicillinase.67 It was
realized that some β-lactams were potential to be substrate mimics that are analogues of the
favourable substrate of the β-lactamases to inhibit the enzymes. In the next 10 years,
significant efforts were made towards the development of more effective β-lactamase
inhibitors. During this period, some semi-synthetic penicillins, such as methicillin and
cloxacillin, demonstrated the capacity to inhibit β-lactamases produced by Gram-negative
29
bacteria. Certain carbapenems (e.g. olivanic acids, pluracidomycins and asparenomycins) were
found to inhibit a wide range of β-lactamases as well.17
Currently, the only β-lactamase inhibitors applied to clinical practice are clavulanic
acid, sulbactam and tazobactam, which are known as mechanism-based β-lactamase inhibitors.
Clavulanic acid was produced by the strain of Streptomyces clavuligerus and discovered in the
1960s, which had potent activity against Class A and certain Class D SBLs. However, it was
inactivated by most Class C SBLs and even worse it had no activity against MBLs.17 The
semi-synthetic sulbactam was discovered in 1978 and initially it was called penicillanic acid
sulfone (PAS). When sulbactam was combined with β-lactam antibiotics (e.g. ampicillin and
cefazolin), they showed an extended broad spectrum activity against both Gram-positive and
Gram-negative pathogens. Although sulbactam could inhibit many SBLs, they were still
inactivated by AmpC (a Class C SBL) and MBLs.17 Later on, tazobactam, a C3'-triazolyl-
substituted PAS, was found to effectively inhibit Class A SBLs, but they underwent reversible
reaction with Class C and Class B SBLs.68
The mechanism for inhibition of Class A SBLs by clavulanate was proposed in the late
1970s (Scheme 8).66 When the acyl enzyme intermediate is generated, a rearrangement occurs
to give the corresponding imine that then isomerizes to a vinylogous amide, which is stabilized
by the conjugation system (tautomer). This stabilized acyl enzyme is not readily hydrolyzed,
leading to inhibition of the β-lactamases.
30
Scheme 8. Inhibition of Class A SBLs by clavulanate
1.4.2 Non-β-Lactam Inhibitors
As early as the 1970s, boronic acid derivatives were reported to have inhibition activity
towards Class A SBL produced by B. cereus. Later on, further studies of the boronic acid
derivatives indicated that they could inhibit S. aureus producing Class A SBLs and
Enterobacteriaceae, and they could efficiently protect ampicillin and ceftazidime.66 However,
they were not used in clinic due to the toxicity of boron. Recently, some more complex
boronic acid were developed by Novartis, which showed inhibition of both SBLs and MBLs.69
Moreover, some phosphonate monoester derivatives were found to be potential β-lactamase
inhibitors as well. It was reported that they could inhibit most SBLs, but they were not
introduced into clinical use either due to their instability in aqueous solution.66
NXL104, a bridged diazabicyclo[3.2.1]octanone developed by Novexel Inc. and
currently in phase II clinical trial, is a non-β-lactam inhibitor with decent activity against SBLs.
31
Combination of ceftazidime and NXL104 in a ratio of 4:1 presents the best activity against
Class A SBLs (mainly TEM and SHV producing pathogens), Class C SBLs producing
Enterobacteriaceae, and Class D SBLs including OXA-48.66
Since the 1980s, a series of cyclobutanones as carbocyclic analogues of β-lactam were
reported by several groups, in which a cyclobutanone moiety replaced the β-lactam ring. They
might be potential inhibitors for both the SBLs and MBLs as well as the PBPs (e.g. D-Ala-D-
Ala transpeptidases). Their synthesis, properties and bioactivities will be discussed in more
detail in the following chapters, which is the main subject of this thesis.
1.4.3 Metallo β-Lactamase Inhibitors
MBLs are the most threatening β-lactamases to humans currently, because they have a
broad spectrum of activity to hydrolyze all β-lactams except the monobactams. The current
clinical β-lactamase inhibitors, such as clavulanate, sulbactam and tazobactam, unfortunately,
are not effectively against MBLs.66 In the last decades, a series of MBL inhibitors were under
development. For example, certain thiol derivatives were found to chelate the zinc ions and
replace the bridging water in the active site of MBLs.66 In addition, pyridine dicarboxylates
and trifluoromethyl ketones were reported to bind to the active site Zn2+ ion within the MBLs
as well.66 Despite of the hydrolysis of clinically used carbapenems by MBLs, some of them
with a variety of side chains at C2 have demonstrated potent inhibition of MBLs. For example,
carbapenem J-110,441 shows good inhibition of IMP-1. Additionally, J-111,225 has activity
against not only IMP-1, but also CcrA, L1 and BcII.66 Succinate derivatives such as 2,3-(S,S)-
disubstituted succinic acid and tricyclic natural products (e.g. SB238569) also show inhibition
of IMP-1.66
32
33
Chapter 2 Previous Work on Cyclobutanone Analogues of β-Lactam Antibiotics
The cyclobutanone analogues of β-lactam antibiotics are compounds in which the β-
lactam nitrogen is replaced by an sp3 carbon. They possess structures and conformations that
are similar to those of the β-lactams, and thus are potential β-lactamase inhibitors to protect
the β-lactams. In addition, such cyclobutanone analogues may act directly, even in the absence
of a β-lactam, as effective antibiotics themselves to eliminate bacteria, if they interact with the
PBPs in a fashion similar to that of β-lactams, again due to their structural and conformational
similarities.
One main advantage of such cyclobutanone mimics, is that they cannot be hydrolyzed
by β-lactamases in the active sites. These cyclobutanones are expected to form a tetrahedral
intermediate, either an enzyme-bound hemiketal in the active site of SBLs or an enzyme-
bound hydrate in that of MBLs (Scheme 9).70 Thus, it is (theoretically) possible to develop a
proper cyclobutanone as a universal and broad-spectrum β-lactamase inhibitor that can inhibit
both SBLs and MBLs at the same time, which would be extremely valuable given the
imminent threat from multidrug-resistant bacteria.
Scheme 9. Cyclobutanones as potential broad-spectrum inhibitors of β-lactamases
2.1 Published Cyclobutanone Mimics of β-Lactam Antibiotics
Starting from early 1980s, several research groups including this group began to study
cyclobutanones as potential β-lactamase inhibitors, and the reported literature examples of
such cyclobutanones are summarized in Figure 5.
34
Figure 5. Reported cyclobutanone mimics of β-lactams in the literature
Cyclobutanone 2.1 made by Gordon and coworkers did not present potent inhibition of
the TEM type of β-lactamase or R61 transpeptidase, but it had some activity against certain
non-β-lactamase producing microorganisms.71 Meth-Cohn's compound 2.2 with an oxime
moiety was proposed to possibly promote acylation of 2.2 in the active site of SBLs, but no
related biochemical data was reported.72 The 2-oxacyclobutanone 2.3 reported by Lowe and
Swain showed time-dependent activity against E. coli R-TEM and B. cereus type I β-
lactamases as well as weak inhibition of R61 transpeptidase.73,74 Compound 2.4 prepared by
Cocuzza and Boswell was a mimic of N-acetyl thienamycin; unfortunately it did not exhibit
antibiotic activity. Interestingly, the benzhydryl esters of 2.4 with the replacement of C3
aminoethylthiolate side chain with sulfoxides and sulfones, showed moderated inhibition of S.
aureus.75 The cyclobutanone 2.5α was synthesized by the Dmitrienko lab, whose synthetic
route and corresponding bioactivity will be discussed in detail in the next section. In the last
decade, cyclobutanone 2.6 was introduced by the Baldwin lab at Oxford as a novel penicillin
mimic.76 Later on the same group synthesized cyclobutanone 2.7 and the corresponding
epimer 2.8 as analogues of penicillin N.77 This group then reported the crystal structure of 2.8
bound to isopenicillin N synthase, which suggested that cyclobutanone 2.8 is a hydrolytically
35
stable mimic of a penicillin (Figure 6).78
Figure 6. Cyclobutanone mimics of penicillin N
2.2 Synthesis of 2-Thiabicyclo[3.2.0]heptan-6-one-4-carboxylate 2.5α 17
In the mid-1980s, a feasible synthetic route towards the cyclobutanone 2.5α was
developed by the Dmitrienko lab at the University of Waterloo, and the corresponding
retrosysntheis is showed below in Scheme 10.
Scheme 10. Retrosynthesis of cyclobutanone 2.5α
The target dichloroethyl ester 2.5α could be traced back to a [2+2] cycloaddition of the
dichloroketene with a decojugated dihydrothiophene ethyl ester 2.9, and the latter could be
obtained from a novel deconjugation of the corresponding conjugated dihydrothiophene acid
2.10. Hydrolysis of ester 2.11 should give the desired acid. Compound 2.11 may be
retrospected to a vinyl phosphonate 2.12 by way of a Horner-Wadsworth-Emmons cyclization
with a mercaptoaldehyde. After a double bond addition step, compound 2.12 could be sought
back to a phosphonate 2.13, which might be furnished from commercially available ethyl
bromoacetate 2.14 and triethyl phosphite 2.15 through a Michaelis-Arbuzov reaction.
As shown in Scheme 11, the lone pair electrons on the phosphorus of triethyl phosphite
36
2.15 attacks the α-carbon of the ethyl bromoacetate 2.14 to displace the bromide and generate
the phosphonium intermediate 2.16, which then is attacked by the bromide ion to furnish the
target phosphonate 2.13 in 98% yield.79 The phosphonate 2.13 is deprotonated by piperidine,
and the anion then undergoes an aldol condensation with paraformaldehyde in methanol to
yield the intermediate alcohol 2.17, which undergoes a dehydration upon treatment with p-
toluenesulfonic acid in refluxing toluene using a Dean-Stark trap to provide the desired vinyl
phosphonate 2.12 (91% yield, 2 steps).
Scheme 11. Preparation of the vinyl phosphonate 2.12
The strategy to synthesize the dihydrothiophene 2.11 was developed by McIntosh and
Sieler (Scheme 12).80 In the presence of TEA, vinyl phosphonate 2.12 reacts with 2,5-
dihydrothiophene 2.18 through a 1,4-addition to give 2.19, which then undergoes an
intramolecular Horner-Wadsworth-Emmons olefination to generate the desired compound 2.11
in 75% yield. Hydrolysis of the ethyl ester 2.11 in aqueous sodium hydroxide affords the
corresponding conjugated acid 2.10 in 85% yield, accompanied by a small amount of a non-
conjugated acid 2.20.
Scheme 12. Preparation of the conjugated acid 2.10
37
As presented in Scheme 13, the conjugated dihydrothiophene 2.10 firstly reacts with
ethyl chloroformate to offer intermediate 2.21. Deprotonation of carbon α to the sulphur of
2.21 triggers the release of carbon dioxide and ethoxide, leading to ketene 2.22. Then the
ethoxide attacks the ketene carbon of 2.22 to form the intermediate 2.23, which is then
protonated to yield the target ester 2.9 (90-94%) that is the precursor for the [2+2]
cycloaddition.81 Two possible side products might be generated in small amounts, including
the conjugated ethyl ester 2.24 (3-7%) and the thiophene ester 2.25 (1-4 %).81
Scheme 13. Deconjugation of the conjugated acid 2.10
The [2+2] cycloaddition of the non-conjugated dihydrothiophene 2.9 with
dichloroketene (generated in situ from dichloroacetyl chloride and TEA) in a non-polar solvent
gives the target dichlorocyclobutanone 2.5α (Scheme 14). This step is somewhat tricky and a
lot of effort has been made to increase the yield of 2.5α. Early studies carried by Darryl
Evanoff of the Dmitrienko lab indicated that slow addition of dichloroacetyl chloride
(dropwise in 2 hours) to TEA and 2.9 in CCl4 could bring the yield up to 40%.82 However,
CCl4 is rather expensive and rare to find nowadays due to its environmental impact. Later on, a
modified strategy was developed by Jarrod Johnson of this group, in which the CCl4 was
replaced by hexane or cyclohexane, and dichloroacetyl chloride (2.5 eq) was added dropwise
over 3 hours through a motor-driven syringe pump to a solution of TEA (2.5 eq) and the
starting material 2.9 in hexane (0.1 M) at room temperature (r.t.), and this solution then was
stirred for additional 21 hours. The yield of 2.5α was improved to 65% by this modified
38
condition and the C4-epimer 2.5β might also be obtained in up to 5% yield, as well as the side
products 2.24 and 2.25 in small proportion. Johnson also found that the maximum practical
scale of this reaction was 75 mmol and the cyclobutanone 2.5α was stable when stored at low
temperature.17
S
CO2Et
Cl2CHCOCl + Et3N
C OCl2C
Hexane
S
CO2EtH
H
O
ClCl S
CO2EtH
H
O
ClCl
+
2.9 2.5 (50-65%) 2.5 (0-5%)
S
CO2Et
S
CO2Et
+
2.25 (5-10%)
+4
1 2356
7
2.24 (5-10%)
Scheme 14. [2+2] Cycloaddition of the non-conjugated dihydrothiophene 2.9 and dichloroketene 2.3 Conformational Studies of Cyclobutanones 2.5
In principle, the five membered ring of the 2-thiabicyclo[3.2.0]heptan-6-ones might
adopt either an endoenvelope or exo envelope conformation (see Figure 7) . X-ray studies by
Lange in the 1980s and later by Johnson of the Dmitrienko group indicated that, in the solid
state, the endo envelope conformation is preferred (Figure 7).83 In the 1H-NMR spectra of
2.5α no coupling is observed between H4 and H5. This observation, coupled with
consideration of the Karplus relationship, suggests that the dihedral angle between H4 and H5
must be close to 90°.84 Such a dihedral angle is expected for the endo envelope conformation
in 2.5α showing that this system has the same conformational preference in solution as it does
in the solid state. In addition, the 4β-CO2Et 2.5β was isolated by Johnson and found to favor
an endo envelope in solution as well. Later on, these results were further confirmed by X-ray
crystallography studies (Figure 7).17
39
Figure 7. Conformational analysis of cyclobutanones 2.5α and 2.5β through molecular modeling and the corresponding X-ray crystal structures 2.4 Derivatization at C3 of Cyclobutanones
Since the mid-2000s, Johnson has synthesized a series of cyclobutanone C3 derivatives
that are potential β-lactamase inhibitors.
2.4.1 Chlorination at C3 of Cyclobutanones
The derivatization started with chlorination at C3 of cyclobutanone 2.5α, which was
treated with SO2Cl2 in CH2Cl2 to give a single chlorinated product 2.26α in near quantitative
yield (Scheme 15). To explain the observed stereoselectivity, it was postulated that in the first
stage, an S-chlorosulfonium ion 2.27 is generated through the initial approach of the
chlorinating agent from the exo face of 2.5α, followed by elimination of HCl, to give the
sulfur-stabilized carbocation 2.28. The leaving group chloride then attacks C3 from the exo
face, leading to the formation of product 2.26α. In the 1H-NMR spectrum, the coupling
constant between H4 and H5 of 2.26α was found to be 6 Hz. As mentioned in Section 2.3, no
coupling between H4 and H5 for the cyclobutanone compounds (e.g. 2.5α) with the endo
conformation was observed. Therefore, 2.26α must adopt an exo envelop conformation. This
40
conclusion was later verified by a single crystal X-ray structure of 2.26α.17
Scheme 15. Chlorination of cyclobutanone 2.5α
2.4.2 Substitutions at C3 of Cyclobutanones
Reaction of 2.26α in water (Scheme 16) gave a mixture of three products, the
substitution product 2.29α and its isomer 2.29β as well as a tricyclic hemiketal 2.29c in a ratio
of 6 : 88 : 6.17
Scheme 16. Reaction between 2.26α and H2O
Furthermore, a series of solvolysis experiments of 2.26α in various polar protic
solvents were carried out by Johnson to generate the corresponding 3-alkoxy derivatives 2.30-
2.34, which are summarized in Table 1.17
41
Table 1. Substitutions at C3 of 2.26α with alcohols and acetic acid
Solvent(s) Time Product Yield (%) OR α (%) β 2.26β 2.35
MeCN/H2O (1:1) 48 h 2.29 75 OH 6 88 0 0
MeCN/MeOH (1:1) 48 h 2.30 73 OMe 75 24 0 1
MeCN/iPrOH (1:1) 40 h 2.31 62 O-iPr 46 20 16 18
MeCN/tBuOH (1:1) 48 h 2.32 60 O-tBu 34 7 44 15
AcOH 48 h 2.33 70 OAc 3 52 43 2
CF3CH2OH 48 h 2.34 64 OTFE 5 76 0 9
It was found that water and small alcohols provide higher yields of the corresponding α
and β isomers, while bulky R (e.g. iPr, tBu, Ac) tend to give lower yields and favour the β
isomer as well as the elimination product 2.35. Moreover, the α isomers are the major products
when the alcohols are MeOH, iPrOH and tBuOH, whereas the β isomers are the primary
products when 2.26α react with acetic acid and trifluoroethanol (TFEOH). However, the β
stereoisomers were found to be unstable on silica gel during purification. Johnson also
proposed a possible mechanism (Scheme 17) involving several intermediates to explain these
observations.17
42
Scheme 17. Proposed mechanism for the steoroselective substitutions of 2.26α by alcohols
On one hand, when the R groups are Me, iPr and tBu, 2.26α is going to lose the
chlorine to form the carbocation 2.36, of which the alcohols could attack from either the exo or
endo face to provide the corresponding α or β isomers, respectively. Moreover, 2.26α could
form a hemiketal 2.37 with the alcohols, which then converts to intermediate 2.38 that prefers
attack from the exo face due to the steric hindrance in the endo face, leading to the formation
of α isomers. Furthermore, 2.37 could form a tricyclic hemiketal 2.40 via intermediate 2.39,
which would also cause the generation of α isomers that are dominant under this situation. On
the other hand, the more acidic solvents such as acetic acid and TFEOH are weaker
nucleophiles and the corresponding reaction pathway is more likely to undergo a simple SN1
process rather than the hemiketal formation (Scheme 18).17 The transition state 2.41 is
generated from an exo (α) face attack of the five-membered ring of intermediate 2.36, but it is
disfavoured due to steric interactions. In contrast, the endo (β) face attack of the
tetrahydrothiophene gives transition state 2.42, which is relatively more stable than 2.41,
providing the β isomers.
43
Scheme 18. Possible mechanism for preferential β selectivity in solvolysis of 2.26α with AcOH and TFEOH A modified condition using silver triflate as catalyst for substitution at C3 was
examined by Johnson as well, which remarkably decreases the reaction time but the yields are
lower than those under the original conditions. Interestingly, the stereoselectivity highly
favored the β isomers in this case.17
The 1H-NMR evidence of these C3 substituted products suggested that cyclobutanones
with 3α substituents (Z = Cl, O-alkyl) prefer the exo conformation while the counterparts with
3β substituents (Z = H, Cl, OAc, O-alkyl) favour the endo envelope (Figure 8). It was
proposed that the J values of H3, H4 and H4, H5 in the endo envelop should be zero, as the
two dihedral angels are around 90°, while those in the exo envelop should not. According to
the corresponding 1H-NMR spectra, in the endo structure, both H3α and H4 appeared as
singlets and H5 appeared as a doublet, whereas, in the exo conformation, a more complex
splitting pattern was observed. The coupling of H3, H4 and H4, H5 could be detected clearly,
which was consistent with the prediction. Later on, these were verified by the corresponding
X-ray crystal structures.17
44
S
H H
H
Z
H
O
Cl
ClO
S
H H
Cl
ClO Z
HH
OEt
Z=Cl, ORexoenvelope
OOEt
Z=Cl, OAc, ORendoenvelope
345
34
5
Figure 8. Conformational preference of C3-substituted cyclobutanone derivatives
In addition, substitution of the C3-chlorine of the ethyl ester 2.26α by thiols (iPrSH and
p-TolSH) and an allyl group (allylTMS) gave the corresponding products, which all strongly
favor the endo envelop conformation.
2.4.3 Elimination at C3/C4 of Cyclobutanones
As illustrated in Scheme 19, the unsaturated ethyl ester 2.35 could be obtained in 82%
yield through an elimination of 2.26α by silver triflate in refluxing dichloromethane. Heating a
solution of 2.26α in dichloromethane with MsOH (10%) at reflux also provides the
unsaturated cyclobutanone 2.35 (72% yield).17 In addition, 2.35 could be prepared by heating
2.26α with TsOH in toluene under reflux, giving the target 2.35 in 94% yield.
Scheme 19. Elimination of cyclobutanone 2.26α
Although the ethyl ester 2.5α can be readily hydrolyzed by 6 M HCl to the free acid
2.43, those analogues with C3 substituents and C3/C4 unsaturation decomposed when
cleavage of the ester ethyl group was attempted under the same conditions. Since the free acid
forms of these cyclobutanones are better mimics of penicillins or penems, a new efficient
strategy was greatly needed.
45
The carboxylic acid 2.43 was prepared from hydrolysis of 2.5α with 6M HCl in
dioxane (79%), which then reacted with Ph2CN2 to yield the benzhydryl ester 2.44 in near
quantitative yield (Scheme 20). Chlorination with N-chlorosuccinimide (NCS), followed by
methanolysis, provided the C3-OMe benzhydryl esters 2.46α and 2.46β in 25% and 5% yield.
Then cleavage of the benzhydryl groups of 2.46α and 2.46β by trifluoroacetic acid (TFA) gave
the corresponding free acids 2.47α and 2.47β in 79% and 61% yield respectively. Eventually,
elimination of 2.47 with 10% MsOH afforded the unsaturated free acid 2.48 in 67% yield.17
Scheme 20. Preparation of C3-methoxy cyclobutanone 2.47 (mimic of penicillins) and unsaturated cyclobutanone 2.48 (mimic of penems) Later on, an improved method for preparing the unsaturated acid 2.48 with much
higher yield (93%) was developed in this lab by Johnson (Scheme 21). The acid 2.43 was
converted into an acid chloride 2.49 with SOCl2, which then was chlorinated by SO2Cl2 to
give the C3-chlorinated acid chloride 2.50. A one-pot elimination and hydrolysis with 10%
MsOH gave the desired acid 2.48.17
Scheme 21. Improved method for synthesizing cyclobutanone 2.48
46
2.5 Reactions at C7 of Cyclobutanones
Some preliminary modifications at C7 of cyclobutanones were also carried out by
Johnson to achieve more cyclobutanone mimics of carbapenems as β-lactamase inhibitors.
2.5.1 C7-Didechlorination of Cyclobutanones
The didechlorinated cyclobutanone 2.51 was obtained in 86% yield upon treating the
dichloroethyl ester 2.5α with Zn/AcOH. The free acid 2.43 and the unsaturated ones 2.35 and
2.48 could be dechlorinated using similar conditions to give the corresponding didechlorinated
Scheme 22. Didechlorination of cyclobutanones 2.5α, 2.35, 2.43, and 2.48
2.5.2 C7-Monodechlorination of Cyclobutanones
Johnson performed some preliminary studies on the monodechlorination of the
cyclobutanones 2.5α (Scheme 23), which reacted with Zn and TMSCl in acetonitrile at 40 °C
for two hours to give the TMS enol ether 2.55. The enol ether underwent aqueous work up to
provide the monochlorocyclobutanone 2.56β in 70-90% yield and the didechlorinated
compound 2.51 in 10-30% yield. However, no monochloro product 2.56α was obtained.
Following the same procedure, the dichlorobenzhydryl ester 2.44 was converted into the
corresponding monochloro product 2.57β (< 40%) and the free acid 2.58β (17-25%), which
are the precursors for the C7-hydroxymethyl cyclobutanones that are analogues of
47
carbapenems. Some attempts had been made to generate the monochloroethyl ester 2.56
directly through the [2+2] cycloaddition between chloroketene and the deconjugated
dihydrothiophene 2.9; however, these reactions were unsuccessful.17
Scheme 23. Monodechlorination of ethyl ester 2.5α and benzhydryl ester 2.44
2.5.3 C7-Hydroxymethylation of Cyclobutanone Derivatives
With the successful monodechlorination, hydroxymethylation at C7 was carried out
sequentially. As shown in Scheme 24, the mixture of 2.56 was treated with TEA in MeCN to
give an enolate, which then underwent an aldol condensation with paraformaldehyde to
generate the hydroxymethyl diastereomers 2.59α (< 13%) and 2.59β (65%). Unfortunately,
they could not be hydrolyzed to the free acid under basic conditions. The C7-hydroxymethyl
acid 2.60β (41-59%) could be obtained from the benzhydryl ester 2.44 via the monochloro
intermediate 2.58 (17%) through a similar monodechlorination-aldol condensation process.17
Scheme 24. C7-hydroxymethylation of cyclobutanone derivatives
48
2.6 Formation of Hemiketals of Cyclobutanones
It was reported by Evanoff and Johnson that cyclobutanones could generate the
corresponding hydrates in D2O (more discussion in Section 2.7) and hemiketals in MeOH-d4,
which are summarized in Table 2 below.17, 82
Table 2. Hemiketal formation of cyclobutanone derivatives
Cyclobutanones α : β Hemiketal Ratio Hemiketal (%)
2.43: R = H 2.5α: R = Et
2.7: 1 2.7: 1
88 91
2.51: R = Et 2.52: R = H
1.8:1 1.6:1
19 24
2.35: X = Cl 2.53: X = H
1.8:1 1.5:1
96 38
S
O
Cl
H
HCl
CO2Et
Z
2.30α: Z = OMe 2.31α: Z = Oi-Pr 2.32α: Z = Ot-Bu
1.2:1 1.1:1 1.1:1
98 98 98
2.30β: Z = OMe 2.31β: Z = Oi-Pr 2.32β: Z = Ot-Bu 2.33β: Z = OAc
4.7:1 4.2:1 1.8:1 1.5:1
15 24 40 30
These results indicate that the chlorocyclobutanones form a larger proportion of
hemiketal in equilibrium with the corresponding keto form. For instance, the dichloro species
2.5α generates 91% hemiketal, while its non-chlorinated counterpart 2.51 generates only 19%
49
of the hemiketal at equilibrium. The electronegative chlorine atoms significantly destabilize
the keto form that has substantial partial positive charge on the carbon of the C=O bond. The
partial positive charge on the corresponding carbon atom in the hemiketal is smaller than that
in the keto form so that the ketal is less destabilized by the halogens than is the ketone.85 For
the C3 derivatives, the cyclobutanones with 3α substituents undergo greater hemiketal
generation (98%), whereas those with 3β substituents form the hemiketals only to an extent of
15% to 40%.
Generally, all these tested cyclobutanones prefer generation of α-hemiketals, and the
hemiketals produced by C3α-substituted compounds favor the exo envelope conformation
while the hemiketals from C3-unsubstituted cyclobutanones prefer the endo envelope (Figure
9). Moreover, the α-hemiketals generated by C3β-substituted cyclobutanones have a tendency
to adopt the endo conformation while the β-hemiketals adopt the exo envelope in order to
avoid the steric hindrance between the OMe group and the substituents in the endo face of the
bicyclic ring.17
Figure 9. Conformational preference of cyclobutanone hemiketals
2.7 Bioactivities (IC50) of Cyclobutanones as β-Lactamase Inhibitors
The inhibition of some common β-lactamases by the cyclobutanones prepared by
Johnson has been tested by Ms. Valerie Goodfellow and Dr. Laura Marrone of the Dmitrienko
group. The inhibition of the chosen β-lactamases by such cyclobutanones are summarized in
Table 3 below.17
50
Table 3. IC50 (μM) of cyclobutanone mimics against common β-lactamases
Ketone inhibitor
% hydrate in D2O
Class A KPC-2
Class B IMP-1
Class B VIM-2
Class C GC1
Class D OXA-10
2.43 74 76 8 1000 1000 25 3 268 8
2.52 0 117 13 235 14 1000 44 3 1135 33
2.30α 98 58 2 122 5 363 9 6.5 1.4 156 6
2.30β 6 99 5 N/A N/A 38 4 547 19
2.35 93 26 2 213 21 1000 4.5 0.3 370 15
2.54 2 170 2 500 N/A 34 3 1000
2.58β N/A 500 260 500 500 500
According to the table, the β-lactamases KPC-2, IMP-1, GC1 and OXA-10 are mostly
inhibited by cyclobutanones that are able to form a larger amount of hydrate (2.43, 2.30α and
2.35). Among these β-lactamases, the SBLs are more efficiently inactivated by the
cyclobutanones while the MBLs are only moderately inhibited by them. Moreover, compound
2.35 presents slightly higher activity against KPC-2 and GC1, which indicates that the
stereochemistry at C4 might significantly affect the binding to the Class A and C SBLs. In
addition, the C3α-OMe cyclobutanone 2.30α shows the best inhibition towards these β-
lactamases, because it was found to have a very similar structure to penicillin with an exo
envelope conformation, which possesses higher affinity for β-lactamases. Later on, X-ray
crystallographic studies provide the evidence that compound 2.30α generate a hemiketal
binding to the serine residue in the active site of OXA-10 (Figure 10).86
51
Figure 10. Interactions of cyclobutanone 2.30α with serine residues in the active site of OXA-10 (Permission was obtained from publisher. See Appendix D for more details.) The former members of the Dmitrienko lab made great contributions to the research on
cyclobutanones. They developed some feasible conditions to prepare the key intermediate, the
dichlorocyclobutanone ethyl ester 2.5α, which was the precursor for the series of
cyclobutanone derivatives, and computational and conformational studies also aided in the
design newer cyclobutanone analogues as β-lactamase inhibitors. To pursue more active β-
lactamase inhibitors, the cyclobutanone project in the Dmitrienko group is continued in this
thesis work, which explores further chemistry at the C7 carbon of this system. More details
will be discussed in the following chapters.
52
Chapter 3 Novel Cyclobutanone Mimics of β-Lactam Antibiotics: Synthesis and
Properties
The mechanism for hydrolysis of β-lactam antibiotics by SBLs involves nucleophilic
attack at the carbonyl carbon of the β-lactam by the hydroxyl group of a serine residue in the
enzyme active site, which leads to the β-lactam ring-opening to form an acyl enzyme
intermediate via a tetrahedral intermediate. In the next deacylation step, this ester is
hydrolyzed by a water molecule that is bound nearby in the active site and activated by a
glutamate residue that acts as a general base. For the MBLs, the hydrolytic mechanism
involves the generation of the tetrahedral intermediate or transition state through nucleophilic
attack of the carbonyl by the zinc bound water of hydroxide (Scheme 25).
Scheme 25. Hydrolysis of penicillin by β-lactamases
In terms of a cyclobutanone mimic, when it is bound to SBLs in the active site, it is
expected to generate a hemiketal adduct, which would be sufficiently long-lived to prevent the
β-lactams from being hydrolyzed. The cyclobutanone analogue is postulated to bind to the zinc
ion(s) in the MBLs active site, and it is predicted to form a hydrate when it is attacked by the
water molecule or hydroxide. Even though some cyclobutanone mimics prepared by the
53
Dmitrienko group showed moderate inhibition of certain MBLs, the corresponding X-ray
crystallographic evidence indicating how the cyclobutanone mimics bind to the MBLs is still
not available at the moment. The proposed mechanisms for cyclobutanone mimics inhibiting
the β-lactamases are illustrated in Scheme 9 of Chapter 2.
3.1 Initial Cyclobutanone Targets
This thesis project initially aimed at synthetically elaborating the cyclobutanones that
more closely resemble meropenem, which is one of the clinically used and last line of defence
carbapenem type of β-lactam antibiotics. The other two major carbapenems currently in
clinical use are imipenem and doripenem (Section 1.1.3). These carbapenems possess a broad
spectrum of activity against both Gram-positive and Gram-negative pathogens, and they are
also found to be resistant to hydrolysis by many SBLs. It has been discovered by several
research groups that carbapenems interact with the active site serine residues of most SBLs to
form an acyl intermediate that is relatively long-lived rather than hydrolyzed rapidly, when
compared to β-lactams without this hydroxyethyl side chain. A crucial structural feature within
the carbapenems is the hydroxyethyl side chain attached to the α position of the β-lactam
carbonyl, which plays an essential role in their resistance to hydrolysis by the SBLs. This is
confirmed by experimental evidence from the Bonomo group (Figure 11). After the generation
of the acyl enzyme intermediate, this hydroxyl group interacts with a water molecule in the
active site through H-bonding, preventing the water from acting as a nucleophile to attack the
ester group of the acyl enzyme.27
54
N
CO2H
S
O
HHHO HN
NH
Imipenem
Figure 11. The hydroxyl group within the side chain of imipenem forms a hydrogen bond with the water molecule in the active site of TEM-1. (Permission was obtained from publisher. See appendix E for more details.) Due to the evolution of bacteria, more recently, a number of SBLs including KPC-2,
OXA-23 and OXA-48 were newly identified as carbapanemases and found to be able to
catalyze the rate of hydrolysis of the carbapenems. Consequently, they threaten almost all the
current clinical β-lactam antibiotics.27
The possible mechanism of the efficient hydrolysis of the carbapenems by the
carbapanemases was reported by Spencer and co-workers at the University of Bristol in 2011.
Some serine β-lactamases, such as SFC-1 (a Class A SBL), are very efficient in hydrolyzing
carbapenems through the key residues in its active site. The active site of such carbapenemase
is enlarged and induces rotation of the hydroxyl group on the hydroxyethyl side chain of the
carbapenems away from the site where the nucleophilic water molecule is bound, making the
hydroxyl group no longer interact with the active site water molecule by hydrogen bonding. As
a result, the acyl enzymes from carbapenems undergo a rapid hydrolysis to release the free β-
lactamases that totally destroys the β-lactams.87 As demonstrated in Figure 12A, the Ser70 of
an Ala mutant of SFC-1 binds to an unhydrolyzed meropenem. The distance between the water
55
molecule and the hydroxyl group on the side chain is 2.70 Å (d1) and an H-bond is observed.
In addition, the distance between Asn132 and the OH group is 2.95 Å (d2), which also
generates an H-bond. While in the acyl enzyme complex (Figure 12B, a Glu166 Ala mutant), it
is observed that the hydroxyl group of meropenem is rotated by approximately 120° when
compared to the former complex. As a result, the distance d2 is slightly decreased to 2.87 Å
and the H-bond between Asn132 and the OH group remains. However, the distance d1 is
significantly increased to 4.97 Å, which is too far to generate an H-bond between the OH
group and the water molecule. Consequently, the water is able to attack the acyl enzyme to
give the hydrolyzed meropenem and release free enzyme.87
Figure 12. (A) An unhydrolyzed meropenem binding to the active site of Ser70 Ala mutant of SFC-1 (B) The acyl enzyme intermediate of meropenem binding to the active site of Glu166 Ala mutant of SFC-1 (Permission was obtained from publisher. See Appendix F for more details.) Initially, the Dmitrienko group proposed a novel idea to address Spencer's discovery.
For the sake of enhancing the stability of carbapenems in the serine active site of the
carbapenemases, the hydroxyethyl side chain could be modified (Figure 13). More specifically,
a side chain with two hydroxyl groups was planned to replace the single hydroxyl side chain of
carbapenems, leading to the formation of a germinal dihydroxyl groups compound or a hydrate
56
of a carbonyl group, which might solve the problem of rapid hydrolysis of carbapenems by the
carbapenemases. If the modified carbapenem with a germinal diol side chain binds to the
SBLs, even though these enzymes are able to orient one of the OH group of the diols away
from the active site water molecule, there is another hydroxyl group remaining that may still
form an H-bond with the water molecule and thus inhibit the β-lactams.
The initial research goal for this project was to synthesize cyclobutanone mimics of
carbapenem with a geminal diol side chain and some related derivatives that are shown in
Figure 13. Theoretically, the geminal diols are not stable as they can undergo a dehydration
process, reversing to the corresponding ketone or aldehyde. Therefore, a very strong electron
withdrawing group, such as trifluoromethyl group (R2 group), is considered to be introduced
into the side chain, which would favour formation of the hydrate.
Figure 13. The proposed geminal diol analogues of meropenem
It should also be noticed that the presence of a chlorine atom at C7 adjacent to the
carbonyl group would prevent the unfavourable equilibrium illustrated in Scheme 26 shown
below. If the chlorine atom was replaced by a hydrogen atom, as mentioned above, the
geminal compounds would likely tautomerize to yield a conjugated enol that is considered to
be relatively stable. The electron-negative C7 chlorine atom not only helps to stabilize the
hydrate (diol) side chain, but also increases the electrophilicity of the carbonyl carbon within
57
the β-lactam ring, making it more readily form a hemiketal with the active site serine residue.
Scheme 26. The potential equilibrium of the meropenem mimics
3.2 Initially Proposed Synthetic Route to Target Cyclobutanones
At the very beginning, a possible synthetic route towards the target compounds 3.15
and 3.16 was proposed as shown in Scheme 27. This project started with the unsaturated free
acid 2.10 available in the Dmitrienko Lab, which then carried on several steps including
deconjugation, basic hydrolysis, [2+2] cycloaddition, acidic hydrolysis and the installation of
benzhydryl group to give the benzhydryl ester 2.45. These reactions have been discussed in
detail in Section 2.2.
Scheme 27. Initially proposed synthetic route towards novel cyclobutanone analogues of meropenem As shown in Scheme 27, a known monodechlorination condition could convert the
58
dichlorocyclobutanone precursor 2.44 to monochlorocyclobutanones 2.57, which then might
undergo aldol condensation with an aldehyde (e.g. CF3CHO, CH2O and CH3CHO) to yield the
corresponding C7 hydroxyalkyl cyclobutanones 3.1-3.3. In the next step, 3.1 and 3.2 are
expected to be oxidized to ketones 3.4 and 3.5, respectively, which then may undergo a
chlorination with NCS to furnish the C3-derivatives 3.6 and 3.7, followed by elimination of
HCl with either silver triflate or 10% MsOH in refluxing dichloromethane to give the
corresponding unsaturated cyclobutanones 3.8 and 3.9. The C3-chlorination and elimination
method was developed by Johnson, as described in Section 2.4.3. The following step is to
introduce a thioether group to the C3 position of 3.8 and 3.9, which can take advantage of the
chemistry developed in the synthesis of thienamycin analogues by the Beecham
Pharmaceuticals group.88 This proposed strategy involves a base-induced conjugate addition of
a thiol 3.10 to the unsaturated ester 3.8 and 3.9, leading to the generation of 3.11 and 3.12,
followed by an enolate formation and overall dehydrogenation with a hypervalent iodine
reagent, providing the desired ketones 3.13 and 3.14, which are expected to be able to generate
spontaneously the hydrate 3.15 and 3.16 in aqueous solutions.
The thiol precursor 3.10 can be prepared by the reported synthetic approach in Scheme
28.89 The commercially available starting material 3.17 reacts with thionyl chloride to turn the
carboxylic acid into an acid chloride, followed by methanolysis to afford the methyl ester that
then is treated with sodium hydroxide and di-tert-butyl dicarbonate to give compound 3.18. In
the next step, 3.18 is converted into a thiol ester 3.19 with the desire stereochemistry through
reaction with TsCl and potassium thioacetate in sequence. Treatment of the thiol ester 3.19 by
sodium methoxide gives the cyclized compound 3.20, which then undergoes an aminolysis
with dimethylamine to yield the target thiol 3.10.
59
Scheme 28. Preparation of the thiol precursor 3.10
3.3 Preparation of the Benzhydryl Ester 2.44
This project started with the conjugated acid 2.10 that was available in the Dmitrienko
lab. Due to the large demand of the benzhydryl ester 2.44 in this project, it was necessary to go
back to the very beginning, starting with the commercially available triethyl phosphite 2.15
and ethyl bromoacetate 2.14. All the details of the related reactions have been discussed in
Chapter 2 and this section just summarizes these reactions and the corresponding yields
(Scheme 29), which have been carried in this thesis work. In fact, cyclobutanone 2.5 is a
racemic mixture with the indicated relative stereochemistry, and all the reaction were
performed with this mixture. For the sake of simplicity, only one enantiomer is shown in all
the diagrams.
Scheme 29. Preparation of precursor 2.44 in this thesis work
3.4 Preparation of the PMB Ester 3.21
Initially, other protecting groups were considered to replace the benzhydryl group to
protect the free acid 2.43, since the PMB source (PMB-Cl or PMB-I) is much safer,
60
particularly at large scale, to handle than the benzhydryl source (Ph2CN2) that is potentially
explosive. A great deal of effort was made towards the synthesis of the PMB ester 3.21, and
the results of such attempts are summarized in Table 4 below.
Table 4. Efforts towards preparation of PMB ester 3.21
Trial Reagent Solvent Temp. Time Result
1 PMB-Cl, NaHCO390 DMF 45°C 3 days N.R.
2 PMB-I, iPr2NEt DMF r.t. 4 h N.R.
3 PMB-I, iPr2NEt Acetone Reflux 23 h 31%
4 PMB-I, NaHCO3 Acetone Reflux 23 h 34%
5 DCC, DMAP, HOBT, PMB-OH DMF r.t. 20 h Decomp.
6 PMB-I, K2CO3 Acetone Reflux 20 h Decomp.
7 PMB-I, K2CO3 Acetone Reflux 8 h Decomp.
8 Ph2PCl, imidazole, PMB-OH, I291 MeCN Reflux 22 h Decomp.
9 Ph2PCl, imidazole, PMB-OH, I2 MeCN Reflux 4 h <29%
In the first trial, sodium bicarbonate was used as the base in order to deprotonate the
acid, and the carboxylate is a potential nucleophile that could attack the benzyl carbon within
the PMB-Cl to displace the chlorine, yielding the desired PMB ester 3.21. Even when this
reaction was carried out in DMF at 45 °C for 3 days, there was no PMB ester formed, as
indicated by 1H-NMR. One possible reason for the unsuccessful synthesis might be the
decomposition of the dichlorocyclobutanone 2.43 by using sodium bicarbonate as the base. It
was discovered by Evanoff that 2.43 is very sensitive to aqueous sodium carbonate. More
exactly, the cyclobutanone ring of 2.43 can be easily opened in 2.5% aq. Na2CO3 at r.t. within
61
4 minutes or 0 °C within half an hour to generate a dicarboxylate 3.22 that then further
degrade to an aldehyde 3.23 (Scheme 30).82 It is possible that the acid 2.43 is partially ring
opened due to the usage of NaHCO3 even though the reaction mixture is non-aqueous, since
NaHCO3 is not strictly anhydrous.
Scheme 30. Degradation of dichlorocyclobutanone 2.43 in aqueous Na2CO3
Another possibility is that the reactivity of the PMB source (PMB-Cl) is not high
enough. Therefore, for the second trial, the base was changed to diisopropylethylamine
(DIPEA), and the PMB source was changed to PMB-I that was prepared by the Finkelstein
reaction of PMB-Cl and NaI.92 This reaction was carried out at r.t. for 4 hours; however, there
was still no target PMB ester obtained.
When the reaction was carried out in refluxing acetone for 23 hours (Trial 3), the
desired PMB ester 3.21 was isolated by flash chromatography in 31% yield. Later on, sodium
bicarbonate was used as the base again in Trial 4, because its solubility in acetone was not
good, the cyclobutanone ring was expected to be stable under this condition, which eventually
provided the desired 3.21 in 34% yield. The optimization continued. It was proposed that
activation of the carboxylic acid might promote the esterification reaction. For Trial 5, a
dicyclohexylcarbodiimide (DCC) coupling was attempted, involving the starting material
treated with 4-dimethylaminopyridine (DMAP), hydroxybenzotriazole (HOBT) and PMB-OH
in DMF (Scheme 31).93 The acid 2.43 is deprotonated by DMAP to give the carboxylate 3.24
that then reacted with DCC to yield the intermediate 3.25. Followed by nucleophilic attack by
HOBT upon the carboxylate carbonyl within 3.25, intermediate 3.27 should be obtained. The
62
HOBT moiety within 3.27 acts as an excellent leaving group in favour of forming the PMB
ester 3.21 upon PMB-OH's nucleophilic attack. However, this condition did not work and no
desired product was isolated. Since this method did not seem promising, no more experiments
following this strategy were performed.
Scheme 31. Proposed mechanism for the DCC coupling
Other conditions were attempted to reach the goal as well (Trial 6 and 7). In these
experiments, potassium carbonate was used as the base in refluxing acetone for 20 hours and 8
hours, respectively. Unfortunately, the starting material decomposed under such conditions.
The last few attempts were focused on the phosphine-imidazole based coupling
(Scheme 32).91 The intermediate 3.30 is generated in situ from the chlorodiphenylphosphine
3.28 and imidazole 3.29, which then is treated with iodine to yield the phosphonium salt 3.31.
Followed by reaction with two moles of the acid 2.43, the acyloxyphosphonium ion 3.33
should be generated, which then may be attacked by the PMB-OH to give the desired PMB
ester 3.21 along with the side product 3.34. For Trial 8 and 9, when the reaction was run for 22
hours, the starting material decomposed. If the reaction time was reduced to 4 hours, only a
small amount of PMB ester 3.21 (< 29%) was generated. This method was not further
explored because of the poor yield and relatively expensive cost of Ph2PCl.
63
Scheme 32. Proposed mechanism for the phosphine-imidazole based coupling
In summary, the best yield to prepare the PMB ester 3.21 is only 34%, which is far
from ideal. More importantly, the following monodechlorination on the PMB ester 3.21 was
also found to be very problematic after a few attempts. Therefore, the PMB ester 3.21 was
"abandoned" and the benzhydryl ester 2.44 was picked again as the key intermediate for this
project, which was obtained from the acid 2.43 reacting with diphenyldiazomethane in ethyl
acetate. Ph2CN2 is a purple solid prepared by oxidation of benzophenone hydrazone by
mercury oxide in a pressure bottle, and it is usually dissolved in EtOAc and stored in the cold
prior to use.
3.5 Monodechlorination of the Benzhydryl Ester 2.44
Since the monochlorocyclobutanone 2.57 is the essential precursor for the C7
derivatives, developing a convenient and efficient method to largely produce the monochloro
cyclobutanone becomes essential.
3.5.1 Method Based on Zinc and TMSCl
The initial attempts towards monodechlorination of the benzhydryl ester 2.44 were
based on Johnson's strategy that has been discussed in Section 2.5.2, involving treatment of the
starting material with zinc dust, TMSCl and anisole in acetonitrile. During this reaction, the
benzhydryl group was cleaved as a carbocation, which might have an adverse effect on the
desired reaction. Therefore, the initial trapping reagent (anisole) was replaced by a much
64
stronger one, 1,3,5-trimethoxybenzene (TMBz). A side product 3.35 derived from alkylation of
the trapping reagent TMBz by the benzhydryl carbocation was isolated by flash
chromatography and characterized.94 Moreover, several conditions were examined to optimize
the reaction (Table 5).
Table 5. Efforts towards monodechlorination with Zn-TMSCla
Trial Reagents Condition Product Ratiob
1 Zn, TMSCl, TMBz 40 °C for 4 h 2.43/2.58α/2.58β (2:1:3, 29%)c
2 Zn, TMSCl, TMBz Reflux for 4 h 2.43/2.58α/2.58β (4:3:10)d
3 Zn, TMSCl, TMBz Reflux for 3 h, 40 °C for 1h 2.58β (24%)e
4 Zn, TMSCl, TMBz Reflux for 2 h, r.t. for 1 h 2.58α/β (1: 3.5, 40%)f
(a) All reactions were carried on in MeCN and all reaction mixtures contained 3.35. (b) identified by 1H-NMR (c) crude yield (d) crude mixture (e) isolated yield (f) purified by flash chromatography It is clear from the table that the yields of these reactions are noticeably increased when
compared to the one (17%) previously obtained by Johnson, because of the use of a better
trapping reagent. The first experiment followed the original condition (40 °C for 4 h) and
provided a mixture of 2.43, 2.58α and 2.58β in a ratio of 2 : 1 : 3, as indicated by 1H-NMR, in
approximately 29% total yield (2.58α/β). The subsequent trials were carried first out in
refluxing MeCN for a certain time (2-4 h), and then further reacted at 40 °C or room
temperature. However, the best result was a mixture of 2.58α and 2.58β (1 : 3.5), obtained in
40% yield after flash chromatography.
These observations suggested that control of temperature and reaction time were
essential for this reaction. In general, when the reaction was carried out in refluxing
65
acetonitrile for a longer time, the benzhydryl group was cleaved. Heating for a shorter time led
to a higher yield for the reaction. However, the maximal yield of 40% is still rather low.
3.5.2 Method Based on Zinc and Ammonium Chloride
As the Zn-TMSCl monodechlorination could not provide the desired monochloro
product in a satisfactory yield, in the next phase of this project, a literature method using Zn-
NH4Cl to dechlorinate was explored on 2.44,95,96 which was obtained through hydrolysis of the
corresponding benzhydryl ester 2.57.
Attempts on this reaction gave a crude product that contained up to five possible
compounds, including the cleaved products 2.57α/β, the left over starting material 2.44, a
didechlorinated product 3.36 and a ring-opening product 3.37 that previously had been isolated
and characterized in this group by Evanoff.82 The results varied significantly depending on the
condition used (Table 6).
Table 6. Efforts towards monodechlorination with Zn-NH4Cla
Trial Reagents Temp. Time β : α : SM : 3.36 : 3.37b
1 Zn (2 eq), NH4Cl (10 eq) 0 °C 3 h 1 : 0 : 1 : trace : 0.5
3 Zn (10 eq), NH4Cl (10 eq) 0 °C 45 min 1 : 0 : 0.5 : 0.1 : 0.4
4 Zn (20 eq), NH4Cl (10 eq) 0 °C 30 min 1 : 0 : 0.3 : trace : 0.2
5 Zn (40 eq), NH4Cl (10 eq) 0 °C 10 min 1 : 0.3 : 0.3 : 0.15 : 0.15
66
Table 6 (continued). Efforts towards monodechlorination with Zn-NH4Cl
6 Zn (50 eq), NH4Cl (50 eq) 0 °C 5 min 1 : 0 : 3 : 0.25 : 0.7
7 Zn* (10 eq)c, NH4Cl (10 eq) r.t. 3 h 3.36 only
8 Zn* (10 eq), NH4Cl (10 eq) r.t. 1 h 3.36 only
9 Zn* (2.5 eq), NH4Cl (10 eq) 0 °C 10 min 1 : 0 : 5 : 0 : 0.75
10 Zn* (5 eq), NH4Cl (10 eq) 0 °C 20 min 1 : 0 : 2 : 0 : 1.2
11 Zn* (10 eq), NH4Cl (10 eq) 0 °C 13 min 1 : 0 : 2.8 : trace : 0.5
12 Zn* (5 eq), NH4Cl (10 eq) -20 °C 3 h 1 : 0 : 2.7 : trace : 1.4
(a) All reactions were carried out in MeOH and worked up as well as examined by 1H-NMR without further purification. (b) ratio in crude product (c) Zn* activated zinc (see Section 4.1) Generally, the results based on the Zn-NH4Cl monodechlorination were not satisfactory.
Even worse, the ring-opening (RO, 3.37) product seems inevitable. Since NH4Cl is a proton
source, the carbonyl carbon may be activated by protonation, making it more susceptible to be
attacked by the solvent methanol as a nucleophile. As a consequence, the cyclobutanone ring is
opened to give the compound 3.37 that shows a characteristic doublet around 6.15 ppm in 1H-
NMR spectra, corresponding to the proton on the carbon bearing the two chlorine atoms.82 In
addition, the monochloro compound 2.57α was produced only in trials 2 and 5 and only in
small amounts.
For the first trial, the starting material (SM) was reacted with 2 equivalents of zinc dust
and 10 equivalents of NH4Cl at 0 °C for 3 hours. The monochloro compound 2.57β was 40%
of the crude product but approximately only 40% of the SM was converted and 20% RO
product was present in the crude product. The next trial doubled the amount of zinc and was
carried out at room temperature in order to promote the conversion of 2.44. The starting
material in this case was completely consumed, but a significant amount of over reduction
67
product was obtained.
In the following experiments (Trials 3-6), the amount of zinc dust was gradually
increased from 10 to 50 equivalents in order to push the reaction, while the corresponding
reaction times were decreased from 45 to 5 minutes to avoid side reactions. The conversions of
2.44 ranged from 40% to 80%, but the crude products still contained some side products,
which were very difficult to separate by flash chromatography. Next, activated zinc (10 eq)
was used in the next two experiments (Trials 7 and 8). Surprisingly, the reactions were very
clean and provided only the OR product 3.36 within 1 hour. Later on, both the amount of zinc
and reaction time were reduced to prevent over reduction, but the conversions of 2.44 were
still not ideal (maximal 52% in Trial 10). Trial 12 further lowered the temperature to -20 °C
and extended the time to 3 hours, which provided only 47% conversion of the starting material.
It was noticed that even at such low temperature, the amount of RO product still could not be
ignored.
All these results indicate that the Zn-NH4Cl monodechlorination is sensitive to
temperature, the quantity of zinc, and reaction time, causing too much difficulty to optimize.
The amount of NH4Cl, however, seems to not influence the reaction much.
In order to test the solvent effect, another condition was tried involving a mixed
solvent system (acetone : methanol = 50 : 1) and activated zinc (5 eq) as well as ammonium
chloride (5 equivalents), which is not shown in Table 6. This condition provided a poor
conversion of SM, and the crude mixture contained the β isomer, starting material and ring-
opening product in a ratio of 1 : 16 : 2.7. The mixed solvent only consisted of 2% methanol,
which still generated a relatively large proportion of ring-opening product, indicating that this
reaction is extremely sensitive to the solvent, particularly nucleophilic ones.
68
If the various side products in the crude mixture could be minimized, the difficulty to
purify them might be reduced. Based on the analysis of the previous results, preventing the
formation of the ring-opening side product 3.37 seemed possible. As mentioned before, the
carbonyl carbon is activated by NH4Cl, and MeOH is a relatively good nucleophile, leading to
the nucleophilic opening of the cyclobutanone ring. To address this problem, a more bulky
alcohol, isopropanol was considered, which is less likely to attack the carbonyl carbon for
steric reasons. Unfortunately, compound 2.44 is not very soluble in isopropanol. Later on, a
mixture of solvent was used to help dissolve 2.44, which consists of iPrOH and acetone in a
ratio of 6:1 (v/v).
The first attempt using this mixed solvent system with non-activated zinc (5 eq) and
ammonium chloride (5 eq) was carried out at 0 °C for one hour, which gave a crude mixture
with the β isomer 2.57β and the starting material 2.44 in a ratio of 1 : 4.5. As predicted, under
this condition, no ring-opening product was observed in the crude mixture, but the conversion
of the starting material needed to be improved. Longer reaction time (5 h) with reduced
quantities of both Zn and NH4Cl (4 eq) was then attempted, since the extended time might lead
to the formation of over reduction product. This gave a crude mixture containing the β isomer
and SM in a ratio of 1 : 1.5, and only trace amount of ring-opening product was observed.
Although the cyclobutanone ring-opening problem had been minimized, the 40% conversion
percentage of 2.44 was far from ideal.
3.5.3 Method Based on Zinc and Ammonium Formate
In order to decrease the reaction time, a stronger reduction condition, involving zinc
and ammonium formate that could even reduce hydroxylamine, was attempted.97 A series of
experiments has been performed and the results are shown in Table 7.
69
Table 7. Efforts towards monodechlorination with Zn-HCO2NH4a
Trial Reagents Solvent Temp. Time β : α : SM : 3.36 : ROb
1 Zn* (5 eq)c,
HCO2NH4 (5 eq)
iPrOH : Acetone
(6 : 1) 0 °C 1 h 1 : 0 : 7.3 : 5.5 : 1.2
2 Zn* (5 eq),
HCO2NH4 (5 eq)
iPrOH : Acetone
(6 : 1) 0 °C 5 min 1 : 0 : 2 : 0 : 0.4
3 Zn* (5 eq),
HCO2NH4 (2 eq)
iPrOH : Acetone
(6 : 1) 0 °C 30 min 1 : 0 : 1.6 : trace : 0.3
4 Zn* (5 eq),
HCO2NH4 (2 eq) Acetone 0 °C 1 h 1 : 0 : 11 : 0 : 0
(a) All reactions were worked up and examined by 1H-NMR without further purification. (b) ratio in crude mixture (c) Zn* activated zinc (d) crude product from Trial 4 used as the SM
70
At the beginning (Trial 1), 5 equivalents of activated zinc dust and a stoichiometric
amount of ammonium formate were used in a mixture of iPrOH and acetone (6 : 1) at 0 °C for
60 minutes. Only around half of the starting material was consumed, and a large amount of
side products such as the ring-opening and over reduction compounds were generated as well.
For the next attempt, all conditions remained the same except that the reaction time was
reduced to 5 minutes. The conversion of 2.44 did not change much, however, the amount of
side products was significantly decreased, and the over reduction product was not formed.
In order to further eliminate the ring-opening product, pure acetone was used as the
solvent. Even when the reaction was performed at higher temperature (r.t.) and a fairly long
time (5 h), the maximal conversion of 2.44 was less than 13%. However, there were no side
products observed in the crude mixture as predicted.
Due to the observed low reactivity in pure acetone, the solvent was shifted back to a
mixed one (acetone : iPrOH = 25 : 1, Trial 6), but it still provided poor conversion of the
starting material. When the solvent was changed to acetone : MeOH = 50 : 1, the
corresponding results (Trial 7 to 11) were not ideal. Although the highest observed conversion
was 68% (Trial 9), the amount of side products could not be ignored.
In summary, the quantity of zinc dust is not very important for this reaction, while that
of ammonium formate somehow leads to over reduction. Fortunately, through careful control
of the amount of ammonium formate, over reduction seems not a big problem. As for the
solvent, reaction in pure acetone and acetone mixed with iPrOH show poor reactivity, whereas
acetone mixed with MeOH gives better conversion, but leads to a small amount of ring-
opening side product. When the reactions were carried on at lower temperature, the rate is
relatively slow. Even though Trials 9 and 10 offered an acceptable conversion and relatively
71
clean crude product, the results were still not adequate.
3.5.4 Reduction with Zinc and Tetrabutylammonium Bisulfate
Since ammonium chloride and ammonium formate have very limited solubility in
acetone, and even in alcohols, other more organic soluble ammonium salts were considered.
Therefore, tetrabutylammonium bisulfate was chosen to perform the reactions, since it had
very good solubility in organic solvents and the bisulfate was rather acidic, which may
promote the reduction. Several monodechlorination attempts under such conditions are
summarized in Table 8 .
Table 8. Efforts towards monodechlorination with Zn- nBu4NHSO4a
Trial Reagents Solvent Temp. Time β : α : SM : 3.36 : 3.38b
(a) All reactions were worked up and examined by 1H-NMR without further purification. (b) ratio in crude mixutre (c) Zn* activated zinc (d) with some other unidentified products accounting for 40%
72
Since the tetrabutylammonium bisulfate may have higher activity, therefore, only one
equivalent of this salt and unactivated zinc dust (5 eq) was used in the first trial. After 2.5
hours at ambient temperature in acetone, only 15% of the β isomer was observed in the crude
mixture, while 85% of 2.44 was left over. In the next experiment, the amount of nBu4NHSO4
was significantly increased to 6 equivalents. Surprisingly, for the same reaction time as Trial 1,
only starting material was recovered. As observed from Trial 3, 10 equivalents of activated
zinc dust were used and tetrabutylammonium bisulfate was kept at 5 equivalents. The reaction
time was extended to 3.5 hours, but only a very small proportion of the desired β isomer was
formed.
In order to understand which reagents affected the results msot, Experiment 4 kept 10
equivalents of zinc dust and dramatically decreased the ammonium salt to 1 equivalent with
2.44 stirring for 2 hours. However, no significant reaction occurred. In addition, in the next
trial (No. 5), the amount of ammonium salt and reaction time were both doubled, but reaction
still did not occur. When a solvent mixutre of acetone : iPrOH (100 : 1) was used, zinc dust
and ammonium salt were kept 4 equivalents and 2 equivalents, respectively (Trial 6), and the
mixture stirred for 8 hours at ambient temperature, ring-opening product and the β isomer
were formed in a ratio of 1 : 3 and significant amounts of some unidentified products were
observed.
To summarize, the unactivated zinc seems to favour the desired monodechlorination,
while the activated zinc does not. On the other hand, the amount of tetrabutylammonium
bisulfate does not have much influence on the reaction. Most of the reactions under the Zn-
nBu4NHSO4 monodechlorination condition led to low conversion. A possible reason is that,
due to the strong acidity of the bisulfate, it might directly react with the activated zinc dust,
73
resulting in insufficient reagents (either Zn or nBu4NHSO4) to perform the desired
dechlorination. Therefore, decreasing the amount of nBu4NHSO4 relative to activated zinc
may favour the desired reaction (Trial 6). Overall this strategy did not seem promising and was
not investigated further.
As discussed (Section 3.5.2-3.5.4), even though some monodechlorination conditions
were examined in detail, the yield of the desired monochloro cyclobutanone 2.57 is still not
ideal.
3.5.5 Method Based on Zinc and Acetic Acid
It was noticed that the conditions for total dechlorination of the dichloroethyl ester 2.5α
involves zinc (5 eq) and pure acetic acid as the solvent to provide the didechlorinated
derivative 2.51 in 86% yield, as carried out by Johnson previously.17 The reagents are very
simple, but the conditions are somehow harsh (80 °C/5 h). It was expected that modification of
this condition (e.g. decreasing the amount of zinc, temperature or reaction time) might be used
to address the challenge.
The first attempt utilized a mixed solvent of AcOH and acetone in a ratio of 1 : 4 (v/v),
which was supposed to reduce the reactivity of Zn. Moreover, only one equivalent of
unactivated zinc dust was initially added to the reaction at room temperature. No product was
observed by TLC after 30 minutes but some was observed after 3 hours. Therefore, an
additional equivalent of zinc was added to promote the reaction, and the reaction mixture was
stirred for another hour. After worked up, the crude product was found to contain the α-isomer,
β-isomer and starting material in a ratio of 0.4 : 1 : 1.1. It was encouraging to find no over
reduction or ring-opening side products were produced. This condition was further optimized,
which are indicated in Table 9.
74
Table 9. Efforts towards monodechlorination with Zn-AcOHa
Trial Reagentb Solvent Temp. Time β : α : SM : 3.36 : ROc
(a) All reactions were worked up and examined by 1H-NMR without further purification. (b) Two aliquots of zinc were added separately. (c) ratio in crude mixture (d) for the first aliquot (e) for the second aliquot It can be seen from the table that when acetic acid was used instead of the mixed
solvent (acetic acid and acetone), the reaction was promoted. Moreover, there was no ring-
opening product generated at all due to the weaker nucleophilicity of acetic acid. For Trial 1,
the first aliquot of zinc was stirred with 2.44 for 2.5 hours, and then the second aliquot of zinc
was added to the mixture that was further stirred for an hour. The starting material was found
to be completely consumed, affording the desired monochlorocyclobutanone 2.57 and over
reduction side product in a ratio of 11 : 1, and this result was found to be very reproducible.
Some more experiments were performed to examine the influences of the amount of
zinc and reaction time. For Experiment 3 the quantity of zinc dust was 1 equivalent in each
aliquot and the reaction time for the first aliquot of zinc was 2.5 hours, but 30 minutes for the
second aliquot. As a result, only 25% of SM was converted into the β isomer. For Trial 4, the
amount of zinc was doubled for both batches, but the reaction time was reduced, and a result
similar to Trial 1 and 2 was obtained. Additionally, the fifth attempt maintained the condition
75
of Trial 1 expect that the second aliquot of zinc was decreased to 0.5 equivalents.
Consequently, approximately only one third of 2.44 was converted into the desired product.
The amount of the second aliquot of zinc dust and reaction time seem to influence the reaction
outcome severely. If the reaction with the second aliquot of zinc is not long enough, such as in
Trial 3, even with enough quantity of zinc, the majority of 2.44 could not be converted. If there
is an insufficient amount of zinc used for the second aliquot (e.g. Trial 5), even though the
reaction time is long enough, the conversion of the starting material is also not ideal.
The crude mixture mainly contained the desired β isomer, and small amount of α
isomer and a minor amount of the over reduction product. This product mixture could not be
purified by flash chromatography and it was noticed that an isomerization (β → α) happened
during the purification attempt. Fortunately, the crude product was pure enough to be used
directly as the starting material for the next step (hydroxymethylation) and the existence of the
over reduction product did not affect the hydroxymethylation. The α and β isomers gave the
same hydroxymethylation product and the minor over reduction product could then be easily
removed upon chromatography. (More details will be given in Section 3.6.)
The monodechlorination with Zn-AcOH is very clean, but it was discovered that the
quality of dichlorobenzhydryl ester 2.44 was quite essential. Some batches of the benzhydryl
ester 2.44 contained a small amount of impurity (mostly benzhydryl related impurity derived
from diphenyldiazomethane). Flash chromatography had to be performed to strictly purify the
benzhydryl ester 2.44 before the reduction. Moreover, unlike Johnson's monodechlorination
condition (Zn-TMSCl), the benzhydryl protecting group could survive under the Zn-AcOH
condition, which is another advantage of this reaction.
With the monochlorobenzhydryl 2.57β in hand, preparation of its free acid form 2.58β
76
was readily achieved. As shown in Scheme 33, the free acid was furnished in 70% yield
through the cleavage of the benzhydryl group by TFA in dichloromethane. The corresponding
antibacterial activity test and hydrate formation experiment of the obtained cyclobutanone
2.58β will be discussed later (Section 3.10 and 3.11). Compound 2.58β is a new cyclobutanone
mimic of β-lactams, generated in this thesis work.
Scheme 33. Cleavage of benzhydryl group of 2.57β by TFA
The successful development of a monodechlorination method paved the way for the
later installation of a side chain at the C7 position of the cyclobutanone, which is considered
as a milestone in this project.
3.6 C7-Hydroxymethylation of the Monochlorocyclobutanone 2.57
It was reported by Johnson that the didechlorinated cyclobutanones 2.52 and 2.54
exhibited poor inhibition of β-lactamases. It was hoped that the C7 modified derivatives might
show improved inhibition by forming possible hydrogen bonds in the active site of the
enzymes, which was evident from computational modeling (Figure 14).17 Binding of the 7β-
chloro-7α-hydroxymethyl cyclobutanone with IMP-1 (Class B MBL) was estimated by
molecular modeling. The hydroxymethyl side chain may be able to interact with the conserved
zinc-coordinating aspartate residue in the active site. Moreover, in the Class D SBLs (e.g.
OXA-48) active site, the side chain might interact with the carboxylated Lys73 that acts as the
general base in the deacylation process through a favorable H-bond. These hydrogen bonds
might improve the affinity of the cyclobutanone towards the β-lactamases.17
77
S
CO2HH
H
O
Cl
HO
C7-Hydroxymethylcyclobutanone
Figure 14. Modeling for C7-hydroxymethyl cyclobutanone as potential β-lactamase inhibitor binding to IMP-1 and OXA-48 As mentioned earlier (Section 2.5.3), the hydroxymethyl derivative 2.60β was obtained
by Johnson in this group, but the maximal yield was only 10% at the time due to the
unsatisfactory monodechlorination, and compound 2.60β was not tested for its bioactivity or
hydrate formation back then. Fortunately, a much better monodechlorination condition was
developed in the current work. As a result, the following hydroxymethylation was also quite
smooth. The improved synthetic method for producing hydroxymethyl derivative 2.60β has
now been applied on gram scales in the Dmitrienko lab (Scheme 34).
Scheme 34. Preparation of hydroxymethyl cyclobutanone 2.60β in this thesis work
The monochlorocyclobutanone 2.57 was obtained in 80% yield through the Zn-AcOH
reaction, and was then treated with TEA and paraformaldehyde in acetonitrile at 50 °C to
generate the 7α-hydroxymethyl benzhydryl ester 3.2β in 68% that could be efficiently purified
by flash chromatography. After cleavage of the benzhydryl group by TFA in the presence of
anisole, the target 7β-chloro-7α-hydroxymethyl 2.60β was generated. The crude product was
very easily purified by simple trituration with cyclohexane, providing very pure 2.60β in 80%
78
yield.
Hydrate formation and bioactivity of the hydroxymethyl derivative 2.60β were then
studied. More details will be given in Sections 3.10-3.11. Moreover, crystals of 2.60β and
2.58β were successfully obtained and X-ray crystal structures were determined by Dr. Assoud.
It was found that cyclobutanone 2.60β can form a tetramer in the unit cell through H-bonds,
which interact between the hydroxyl group within the C7 side chain and the carboxylic acid at
C4 (Figure 15A). For the cyclobutanone derivatives without the hydroxymethyl side chain at
C7 such as 2.58β, they commonly generate a dimer by the H-bond between the two carboxylic
acids at C4 (Figure 15B). These crystal structures have confirmed the stereochemistry at C7
that was previously assigned by NMR experiments. In these structures, the 5-membered ring
takes up an endo envelope conformation.
Figure 15. (A) Crystal structure of hydroxymethyl cyclobutanone 2.60β (B) Crystal structure of monochlorocyclobutanone 2.58β
3.7 Attempts to Install Trifluoroacetyl Side Chain at C7
Several attempts were made to prepare the trifluoromethyl ketone 3.4 based on an
initial aldol condensation the ketone 2.57 with trifluoroacetaldehyde to be followed by
oxidation of the alcohol (Scheme 35).
79
Scheme 35. Possible synthetic approach to C7-trifluoroacetyl cyclobutanone 3.4
Triflouoroacetaldehyde must be generated in situ from the commerically available
trifluoroacetaldehyde ethyl hemiacetal.98 The first attempt at the aldol condensation used TEA
as the base as in the successful hydroxymethylation reaction for this system. Unfortunately the
crude product that was obtained was found to be very complicated and inseparable.
Several research groups have reported methods for the preparation of
hydroxytrifluoroethylated compounds by using the hemiacetal of CF3CHO such as the
examples shown in Scheme 36.99,100
Scheme 36.(A) Synthesis of the hydroxyltrifluoroethyl ketone through aldol condensation between trifluoroacetaldehyde ethyl hemiacetal and enamines in the Funabiki group (B) Aldol condensation between trifluoroacetaldehyde ethyl hemiacetal 3.39 and cyclohexanone 3.43 in the Gong group An effort was made to apply the strategy published by the Gong group to the synthesis
of 3.3. The proposed mechanism of this reaction with cyclobutanone 2.57 is shown in Scheme
37.100
80
S
CO2CHPh2H
H
O
ClH N
H
S
CO2CHPh2H
HClH
N
HO
S
CO2CHPh2H
HCl
N
- H2O
OH
OEtF3C+ H2O
CF3CHO S
CO2CHPh2H
H
N
Cl
F3C
O
H2OS
CO2CHPh2H
HCl
F3C
OH
O
+NH
2.57 3.45 3.46
3.39
3.47
3.1
Scheme 37. Proposed mechanism for the pyrrolidine-catalyzed aldol condensation between cyclobutanone 2.57 and CF3CHO (generated in situ) Unfortunately, several attempts at this reaction yielded complex mixtures of products
from which none of the desired product could be isolated. Therefore, at this stage, it was
decided that this aspect of the initial proposal would not be pursued further. Instead, the
following new tasks were planned to focus on the modification of the hydroxymethyl
cyclobutanone 2.60β at C7.
3.8 Dechlorination of the Hydroxymethyl Cyclobutanone Derivatives
Since the synthesis of the trifluoroacetyl compound 3.4 was not promising, a new
research direction that involved the dechlorination of the hydroxymethyl cyclobutanone 3.2β
was proposed. The potential procedure for the dechlorination could follow that mentioned in
Section 3.5.2 (Zn-NH4Cl), the dechlorination process may involve the enol 3.48 formation,
which should be then protonated to give the dechlorinated compound 3.49. The enol has a
planar conformation, allowing the proton to be added from either side, leading to a mixture of
C7 epimer 3.49α and 3.49β. The dechlorinated compound 3.49 then could be subjected to a
deprotection condition with TFA and anisole to provide the corresponding acids 3.50 (Scheme
38).
81
S
CO2CHPh2H
H
O
Cl
HO
3.2
Zn, NH4Cl
MeOH
S
CO2CHPh2H
H
HO
HO
3.48
TFA, anisole
CH2Cl2
S
CO2HH
H
O
H
HO
3.50
H+ S
CO2CHPh2H
H
O
H
HO+
S
CO2CHPh2H
H
O
H
HO
3.49 3.49
+S
CO2HH
H
O
H
HO
3.50
7
Scheme 38. Possible procedure for preparation of dechlorinated hydroxymethyl cyclobutanone derivatives It was proposed that the hydroxymethyl cyclobutanone 3.50β with the hydroxymethyl
side chain on the β face of the cyclobutanone ring is a potential β-lactamase inhibitor. As
shown in Scheme 39, 3.50β is considered to bind to both SBLs and MBLs, providing the
corresponding tetrahedral intermediates 3.51 and 3.52. The hydroxyl group of the
hydroxymethyl side of 3.50β may interact with the tetrahedral intermediates to form a six-
membered ring through favourable intramolecular hydrogen bond, which might promote the
stability of the intermediates, preventing them from further generating the enzyme
intermediates and releasing the free β-lactamases. The hydroxymethyl group of the
cyclobutanone 3.50β and the alkoxides in the tetrahedral intermediates orient towards the same
side of the cyclobutanone ring, which might allow them to generate an intramolecular H-bond.
The isomer 3.50α might also be a β-lactamase inhibitor since it is a mimic of carbapenem.
However, its inhibition may be weaker than 3.50β, because the hydroxymethyl group and the
alkoxides are on the different sides of the cyclobutanone ring, as a result, the distance and
geometry between them is probably too far to form such intramolecular H-bond.
82
Scheme 39. Potential inhibition of β-lactamases by cyclobutanone 3.50β
The previous exhaustively examined Zn-NH4Cl condition (Section 3.5.2) was chosen
to dechlorinate the cyclobutanone 3.2β. Since there is only one chlorine atom at C7 to be
removed in this case, so there is no over reduction issue as previously encountered. Activated
zinc was used and the corresponding experimental results are summarized in Table 10.
Table 10. Efforts towards dechlorination of the hydroxymethyl cyclobutanone 3.2β
Trial Reagent Solvent Temp. Time α : β : SMb
1 Zn* (5 eq) a, NH4Cl (5 eq) MeOH r.t. 3 h 1 : 1 : 2
(a) ratio in crude mixture identified by 1H-NMR, (b) crude product from Trial 1 used as the SM (c) Too much aromatic impurity (d) crude product from Trial 5 used as the SM The initial trial used 2.4 equivalents of pyridine and 1.2 equivalents of benzoyl
chloride in dichloromethane at r.t. for 2 hours, and the crude 1H-NMR indicated that 60% of
the starting material was left over. Therefore, the second trial increased the reaction time to 21
hours, but there was still about 30% of starting material unreacted. As described in Trial 3,
even when the amount of pyridine and BzCl was doubled, even after 18 hours there was still
20% of unreacted cyclobutanone 3.2β in the crude product.
In the fourth experiment, the amount of pyridine was significantly increased to 20
equivalents and BzCl to 10 equivalents in refluxing dichloromethane for 17 hours. The starting
material was fully converted, but too much aromatic impurity derived from the large excess of
BzCl, which made the purification very difficult. Therefore, in Experiment 5, the amount of
both reagents was decreased and a catalytic quantity of DMAP was added in order to pursue a
complete conversion of 3.2β. However, only 35% of starting material was transformed. The
quantity of pyridine and BzCl was increased to 10 and 5 equivalents, which were stirred with
the starting material (crude product of Trial 5 to give completed conversion for Trial 6). This
crude mixture was cleaner than that produced from Trial 4, and it was possible to purify the
88
compound by flash chromatography. The target compound 3.62β was isolated in 62% yield.
However, Trial 7 directly used 10 equivalents of pyridine and 5 equivalents of BzCl, which
still afforded incomplete conversion of 3.2β. Later on, Trial 8 with proper amounts of pyridine
(15 eq) and BzCl (8 eq) finally provided a full conversion of the starting material, and the
desired product 3.62β was isolated by flash chromatography in 67% yield.
Clearly, 3.2β is a sterically hindered alcohol that requires conditions involving larger
quantities of reagents and higher temperature than those required for acylation of a simple
alcohol. The next reaction was to cleave the benzhydryl group of 3.62β to provide the desired
cyclobutanone acid 3.55β (Scheme 44).
Scheme 44. Preparation of the carboxylic acid 3.55β
After much experimentation, it was eventually found that the best procedure involved
reaction of 3.62β with forty equivalents of TFA and seven equivalents of anisole in CH2Cl2 in
an ice bath for 9 hours. The crude mixture was successfully purified by HPLC, providing
3.55β in an isolated yield of 23%.
3.10 Hydrate Formation of Cyclobutanone Derivatives
The cyclobutanone derivatives are capable of generating hydrates in aqueous solution,
indicating that they might form an enzyme bound hydrate in the active sites of MBLs, which
may inhibit the MBLs. Based on Johnson's method, a series of cyclobutanone derivatives have
been tested in a mixed solvents system (D2O : acetone-d6 = 3 : 1) in order to generate the
corresponding hydrates, which are summarized in Table 12. Theoretically, the hydrate
formation experiment should be carried out in pure D2O, but many cyclobutanones have very
89
low solubility in water. Thus, acetone-d6 was used as a co-solvent to help dissolve the
substrates.
Table 12. Summary of hydrate formation of cyclobutanones prepared in this thesis work
Trial Cyclobutanone Time Hydrate (%)a
1 2.58β 5 min 25
2 2.60β 15 min 12
3 3.50α 5 h < 2b
4 3.50β 489 h < 2
(a) determined by 1H-NMR (b) detection limitation of 500 MHz NMR spectrometer
For monochlorocyclobutanone 2.58β, an equilibrium mixture containing 25% of
hydrate was formed within 5 minutes and this ratio did not change in the next 18 hours. The
7β-chloro-7α-hydroxymethyl cyclobutanone 2.60β took somewhat longer to achieve an
equilibrium mixture containing 12% of the corresponding hydrate. The corresponding 13C-
NMR experiments were carried out as well; the related peaks for the hydrate are reported in
Chapter 4.
It was observed that the signal of the 7α proton within 2.58β disappears gradually (5.57
ppm) during the hydrate formation process. As proposed in Scheme 45 below, this proton
might slowly exchange with deuterium (from D2O).
90
Scheme 45. Possible H/D exchange between 2.58β and D2O
For the 7α-hydroxymethyl derivative 3.50α, no hydrate formation was detected by
NMR under these conditions. In the case of the 7α-benzoyloxymethyl-7β-chloro derivative
3.55β, it was not possible to carry out the NMR experiment because of poor solubility.
In the case of the 7β-hydroxymethyl cyclobutanone 3.50β, some weak signals were
seen in the 1H-NMR spectrum at 8 minutes after addition of D2O to an acetone-d6 solution of
3.50β. However, later it was concluded that these signals at 5.38 ppm and 5.82 ppm likely
arise from the elimination product 3.65 (Scheme 46). The percentage of this side product in
the reaction mixture gradually increased to a maximum of 25% after 126 hours. The chemical
shifts for the vinyl protons of a compound such as 3.65 are predicted to be in the 6.2 to 6.3
ppm range based on empirical calculations.105 After a careful search of the literature, however,
it was found that Wasserman and co-workers had synthesized the parent 2-
methylidenecyclobutanone 3.66 in 1980 and had found that the vinyl hydrogens had chemical
shifts of 4.99 and 5.62 ppm. The reason for the anomalously low chemical shifts is not clear,
they are sufficiently close to those observed in this study to allow us to speculate that they
arise from the elimination product 3.65.106
Scheme 46. Possible elimination of 3.50β in water and acetone
91
In summary, the cyclobutanone derivatives with a chlorine atom at the C7 position are
more likely to undergo hydration, which is consistent with Johnson's previous results (Table 3,
Section 2.7). When compared to the dichlorocyclobutanones 2.43, 2.30α and 2.35, the amount
of hydrates produced by the monochlorocyclobutanones 2.48β and 2.50β are much less,
because of the difference in electrophilicity of the carbonyl carbon that was affected by the
nearby substituents.
3.11 Bioactivity of Cyclobutanone Derivatives as β-lactamase Inhibitors
The inhibition of some common β-lactamases by the cyclobutanone compounds
prepared in this thesis work has been tested by Dr. Geneviève Labbé and Mrs. Valerie
Goodfellow of the Dmitrienko group. The corresponding biological data are shown in Table 13.
Table 13. Inhibition of some common β-lactamases by the cyclobutanones
β-Lactamases Inhibition (%)
500 μM 2.58β 500 μM 3.50β
Class A KPC-2 1 7
Class B IMP-1 59 25
Class B VIM-2 40 0
Class B SPM-1 34 0
Class B L1 48 0
Class B NMD-1 0 0
Class C GC1 41 18
Class D OXA-10 19 0
The inhibition was assayed by Dr. Geneviève Labbé through nitrocefin hydrolysis.
92
In general, the cyclobutanone mimic 3.50β has poor inhibition of these β-lactamases,
even at a rather high concentration (500 μM). Even worse, 3.50β does not show any activity
against Class B and Class D enzymes that were tested, while the cyclobutanone analogue
2.58β demonstrates moderate inhibition of these β-lactamases at 500 μM except KPC-2 and
NMD-1. These observations are consistent with Johnson's previous conclusion (Table 3,
Section 2.7) that the cyclobutanone capable of generating the larger amount of hydrate in
aqueous solution is more likely to be a better β-lactamase inhibitor.
Some antimicrobial assays of cyclobutanones 2.58β, 2.60β and 3.50β as β-lactam
mimics were also carried out in this group to determine their MIC (Minimum Inhibitory
Concentration), which is the lowest concentration of the drug that results in no growth of the
bacterium. As seen in tables 14, none of the cyclobutanone compounds showed antibacterial
activity on its own (MIC > 256 μg/mL or > 128 μg/mL) with any of the highly resistant
clinical isolates tested. Each of the clinical isolates examined is resistant to the carbapenem,
meropenem (MIC >> 32 μg/mL). The possibility that the cyclobutanones might act
synergistically with meropenem was tested in the checkerboard synergy experiments. In these
experiments the MIC for meropenem was determined at a series of concentrations of the
cyclobutanone (0, 8, 16, 32, 64, 128 and 256 μg/mL). Only in the case of 2.58β with a
meropenem-resistant strain of Stenotrophomonas maltophilia was synergy observed, where the
meropenem's MIC dropped to 32 μg/mL at a concentration of 32 μg/mL of the cyclobutanone,
as compared with 128 μg/mL for meropenem in the absence of the cyclobutanone.
93
Table 14A. MIC values for meropenem in the presence of cyclobutanone 2.58β
Crystal Data and Structure Refinement for Cyclobutanone 2.60β
Empirical formula C8H16ClO4S Formula weight 236.66 Temperature 200(2) K Wavelength 0.71073 Å Crystal system Triclinic Space group P-1 Unit cell dimensions a = 5.31710(10) Å, b = 8.9724(2) Å, c = 10.5459(2) Å α = 78.3564(12)°, β = 77.0590(10)°, γ = 77.6125(10)° Volume, Z 472.630(17) Å3, 4 Density (calculated) 1.663 Mg/m3 Absorption coefficient 0.608 mm-1 F(000) 244 Crystal size 0.200 × 0.080 × 0.020 mm3 Theta range for data collection 2.008 to 26.368° Index ranges -6<=h<=6, -11<=k<=11, -13<=l<=13 Reflections collected 7781 Independent reflections 1933 [R(int) = 0.0221] Completeness to theta = 25.000° 99.9 % Absorption correction Semi-empirical from equivalents Max. and min. transmission 0.7460 and 0.6923 Refinement method Full-matrix least-squares on F2 Data / restraints / parameters 1933 / 37 / 135
Goodness-of-fit on F2 1.038 Final R indices [I>2sigma(I)] R1 = 0.0318, wR2 = 0.0739 R indices (all data) R1 = 0.0413, wR2 = 0.0788 Largest diff. peak and hole 0.386 and -0.181 e.Å-3
119
Atomic Coordinates (× 104) and Equivalent Displacement Parameters (Å2 × 103) for 2.60β
Crystal Data and Structure Refinement for Cyclobutanone 2.58β
C1
S2 C3
C4C5
C6C7
Cl12
C8O9
O10
O11
S
CO2HH
H
O
ClH
2.58
Empirical formula C7H7ClO3S Formula weight 206.64 Temperature 296(2) K Wavelength 0.71073 Å Crystal system Monoclinic Space group P21/c Unit cell dimensions a = 5.93510(10) Å, b = 21.2896(4) Å, c = 7.16860(10) Å α = 90°, β = 107.9793(10)°, γ = 90° Volume, Z 861.56(3) Å3, 4 Density (calculated) 1.593 Mg/m3 Absorption coefficient 0.646 mm-1 F(000) 424 Crystal size 0.170 × 0.080 × 0.020 mm3 Theta range for data collection 1.913 to 27.999° Index ranges -7<=h<=7, -28<=k<=27, -9<=l<=8 Reflections collected 8836 Independent reflections 2075 [R(int) = 0.0205] Completeness to theta = 25.242° 100.0 % Absorption correction Semi-empirical from equivalents Max. and min. transmission 0.7460 and 0.7059 Refinement method Full-matrix least-squares on F2 Data / restraints / parameters 2075 / 0 / 110
Goodness-of-fit on F2 1.207 Final R indices [I>2sigma(I)] R1 = 0.0463, wR2 = 0.1002 R indices (all data) R1 = 0.0685, wR2 = 0.1105 Largest diff. peak and hole 0.390 and -0.351 e.Å-3
122
Atomic Coordinates (× 104) and Equivalent Displacement Parameters (Å2 × 103) for 2.58β
Crystal Data and Structure Refinement for Cyclobutanone 3.50α
C1S2
C3
C4C5
C6
C7
C8O9
O10
C11
O12
O13S
CO2HH
H
O
H
HO
3.50
Empirical formula C8H10O4S Formula weight 202.22 Temperature 296(2) K Wavelength 0.71073 Å Crystal system Monoclinic Space group P21/c Unit cell dimensions a = 5.5026(2) Å, b = 12.7899(6) Å, c = 12.3981(10) Å α = 90°, β = 99.047(2)°, γ = 90° Volume, Z 861.70(6) Å3, 4 Density (calculated) 1.559 Mg/m3 Absorption coefficient 0.353 mm-1 F(000) 424 Crystal size 0.314 × 0.160 × 0.060 mm3 Theta range for data collection 2.303 to 27.986° Index ranges -7<=h<=7, -28<=k<=27, -9<=l<=8 Reflections collected 8652 Independent reflections 2085 [R(int) = 0.0151] Completeness to theta = 25.242° 100.0 % Absorption correction Semi-empirical from equivalents Max. and min. transmission N/A Refinement method Full-matrix least-squares on F2 Data / restraints / parameters 2085 / 0 / 126
Goodness-of-fit on F2 1.081 Final R indices [I>2sigma(I)] R1 = 0.0306, wR2 = 0.0714 R indices (all data) R1 = 0.0360, wR2 = 0.0751 Largest diff. peak and hole 0.292 and -0.221 e.Å-3
125
Atomic Coordinates (× 104) and Equivalent Displacement Parameters (Å2 × 103) for 3.50α